Generative AI regulation targets the development, deployment, and use of large language models, image generators, and other foundation models that create new content. State legislatures are introducing bills covering training data transparency, output labeling, safety testing, deepfake prevention, and liability for AI-generated content.
Key regulatory themes include mandatory disclosure of AI-generated content, watermarking requirements, training data provenance, model safety evaluations, and restrictions on using generative AI for fraud, impersonation, or election interference.
Artificial Intelligence Safety Measures Act
Illinois' SB 315, the Artificial Intelligence Safety Measures Act, regulates developers of frontier AI models (foundation models trained with more than 10^26 operations). All frontier developers must publish a transparency report when deploying a new or substantially modified model and report critical safety incidents within 72 hours. Large frontier developers (over $500 million revenue with affiliates) need a state disclosure statement from January 1, 2027, and from January 1, 2028 a published frontier AI framework and annual independent audits. Large developers face Attorney General penalties up to $1 million, then $3 million. Signed July 6, 2026 (P.A. 104-0538); effective January 1, 2027.
An act relating to data brokers and personal information
Vermont's H.211 (Act 138 of 2026) overhauls Vermont's data broker law, with higher fees and fines, checks before brokers disclose data, and a data broker breach-notice law. Its AI-relevant provision is in the annual registration under 9 V.S.A. § 2446: a data broker must state whether, in the past year, it shared consumers' data with or sold it to a developer of a generative AI ("GenAI") system or model, alongside disclosures about sales to foreign actors, governments and law enforcement. The act defines a GenAI system as an AI system that can generate synthetic text, images, video and audio. These provisions take effect January 1, 2027.
Generative AI Terrorism Risk Assessment Act
The federal H.R. 1736 (Generative AI Terrorism Risk Assessment Act) would require the Secretary of Homeland Security, in consultation with the Director of National Intelligence, to submit an assessment of terrorism threats posed by the use of generative AI within one year after enactment and annually for five years after that. The House passed it by voice vote November 19, 2025; received in the Senate November 20, 2025 and referred to the Homeland Security and Governmental Affairs Committee.
Companion chatbots: crisis interruption pauses.
California's AB 1988, the Preventing AI User Self Endangerment (PAUSE) Act, would have required companion chatbot operators to adopt a public policy for credible crisis expressions (intent to harm self or others), warn the user on the first one, and on a second within 72 hours start a crisis interruption pause that continues until a human moderator reviews the chat. It passed the Assembly 73-0 on May 21, 2026; its three Senate committee hearings (June 12, 22 and 26) were canceled at the author's request, and the bill died. It did not become law.
Requires certification of filings produced using generative artificial intelligence; requires the brief of an appellant to contain a disclosure of the use of generative artificial intelligence in the drafting of the brief and certification that the content therein was reviewed and verified by a…
New York's S 9640 (2023-2024) would have added CPLR rule 2107 requiring any paper served that was drafted with generative AI assistance to attach an affidavit disclosing the use and certifying that a human reviewed the source material and verified the content, and amended CPLR rule 5528 for appellate briefs. Referred to Judiciary May 16, 2024; it died when the 2023-2024 session ended.
Provides that in any civil, criminal or family court proceeding, where evidence is offered and a party contends that such evidence has been fabricated by means of generative artificial intelligence, the court shall not, on that ground alone, conduct an inquiry into such alleged fabrication unless…
New York's S 9390 would add CPLR rule 4540-b, Criminal Procedure Law section 60.80 and a Family Court Act section on evidence a party contends was fabricated with generative AI. A court could not inquire into the alleged fabrication on that ground alone unless the contending party makes a showing sufficient to support a reasonable inference that the evidence may not be authentic. If that showing is made, the evidence is admissible only if the court finds by a preponderance of the evidence that the offering party has established authenticity and relevance. Referred to Judiciary March 6, 2026; it has not passed either chamber.
Data brokers: data collection and deletion.
California's SB 361 expands what data brokers must report when registering with the California Privacy Protection Agency. Registrations must say whether the broker collects data such as account logins, government ID numbers, citizenship or immigration status, union membership, sexual orientation, gender identity or biometric data, and whether in the past year it shared or sold consumers' data to a developer of a generative AI system or model, a foreign actor, governments or law enforcement. Signed October 8, 2025 (Chapter 466); effective January 1, 2026.
Consumer Protection and Product Liability - Chatbots
Maryland's HB 1250 (2026, first reader text, cross-filed with SB 827) would have regulated operators and developers of generative AI chatbots made available to Maryland users: affirmative consent for training on input data, a ban on using chat logs for advertising, a 10-year cap on chat log retention, protections for users under 13, hourly not-human warnings, monthly safety testing, and product liability with possible strict liability. It had a House Economic Matters hearing set for March 3, 2026 and died when the 2026 session ended.
Consumer Protection - Companion Chatbots - Regulation
Maryland's HB 952 (2026, third reader text) would have required operators of companion chatbots available to Maryland users to keep and publish self-harm and minor sexual-content protocols, display warnings that the chatbot is not human (a persistent on-screen warning plus pop-ups at the start and after every hour of continuous use), limit personal data use, and run a complaint system. A chatbot would be treated as a product with strict liability exposure. It passed the House 123-4, was referred to Senate Finance, and died when the 2026 session ended.
Election Law - Influencing a Voter By Dissemination of a Deepfake - Prohibition
Maryland's HB 525 (2025, third reader) would have made it unlawful to influence or attempt to influence a voter's decision, with fraudulent intent, by disseminating a deepfake: a realistic but false depiction of a person (new Election Law section 16-905). A violation would have been a misdemeanor (fine up to $5,000, up to 5 years, or both). It passed the House 127-10 and was referred to the Senate Rules Committee on April 4, 2025, with no further action; it died when the session ended.
Election Law - Campaign Materials - Disclosure of Use of Synthetic Media
Maryland's HB 740 (2025) would have required candidates, campaign finance entities, certain registrants and their agents to include a disclosure when they distribute campaign material made with or containing synthetic media (an image, audio or video intentionally manipulated with generative AI or other digital technology to create a realistic but false depiction of a real individual's appearance, speech or conduct) to anyone in Maryland. Violations would have been a misdemeanor (fine up to $1,000, up to 1 year, or both). A hearing was set for February 11, 2025; it went no further and died when the session ended.
Relating To Artificial Intelligence.
Hawaii's HB 2500 (2026) would have added a chapter on algorithmic decision systems. Developers would have to give deployers risk disclosures; deployers would have to give notice before use and post-decision information (including the 20 most influential characteristics) when a system affects education, employment, finance, government services, health care, housing, insurance or legal services; generative AI would have to be disclosed to users. Developers and deployers would be jointly and severally liable for violations facilitated by the deployer. House committee ECD deferred it February 6, 2026; it died.
Relating To Artificial Intelligence.
Hawaii's HB 639 HD 1 (2025) would have required businesses and individuals in commercial transactions to clearly disclose before using an AI chatbot that could lead a reasonable person to think they are talking to a natural person, and chatbot developers to disclose their chatbot uses AI and can mimic human behavior. It would have allowed suits with the greater of $1,000 or treble damages and civil penalties up to $5,000,000. It passed House second reading February 5, 2025 and was referred to CPC, was carried over to 2026 on December 8, 2025 and was not enacted.
Relating To Artificial Intelligence.
Hawaii's HB 2152 (2024) would have set up a plan for state use of generative AI: Office of Enterprise Technology Services risk assessments and procurement guidelines, an inventory of high-risk uses by each agency, employee training, a duty to tell people when AI mediates electronic communication, and chief information officer approval before adopting an automated decision system. It was referred to HET/LGO, CPC and FIN on January 24, 2024 and was not enacted.
Requesting The Department Of Education To Develop Artificial Intelligence Literacy Curricula Plans And Programs For Use In Public Schools.
Hawaii's Senate Concurrent Resolution 33 (2026) would have requested the Department of Education to develop artificial intelligence literacy curricula plans and programs for public schools, covering AI concepts, machine learning and generative AI, best practices, ethics and AI's relation to Hawaii's history, environment and culture. They were requested to the Legislature 20 days before the 2027 session. It was referred to Senate committees on March 19, 2026 and never adopted. A resolution does not change the law.
Artificial intelligence chatbot technology requirements provided, and cause of action for harm created.
Not law. Minnesota's HF 4452 would have added Minn. Stat. 604.115 barring chatbot proprietors from letting chatbots give substantive medical, mental health or legal advice that would require a license, with a private right of action a notice could not waive, and requiring clear notice that users are talking to an AI. Companion chatbot operators would have had to suspend use for at least 72 hours and show crisis contacts on self-harm, and would be liable if they knew of it and failed to act. Introduced March 18, 2026.
Interactions with Artificial Intelligence
Not law: died in committee. Florida's HB 659 would have created s. 501.172, F.S., requiring operators of companion chatbots to display a notice that companion chatbots are AI-generated, not human and may not be suitable for some minors, maintain a protocol against suicide and self-harm content that refers users to the 988 lifeline, offer age verification and, for known minors, disclose the AI, show a break reminder at least every 3 hours and take reasonable measures against sexually explicit material. Annual reports to the Department of Legal Affairs would have started July 1, 2027. It died in subcommittee on March 13, 2026.
Advanced Technology
Laid on the table; parts became law through its companion. Florida's CS/CS/CS/HB 1459 (2024) would have created the Government Technology Modernization Council, required safety and transparency disclosures for commercial AI-generated chatbot, image, audio and video output (s. 501.174), and made generated child pornography a third-degree felony. Companion CS/CS/SB 1680 (Chapter 2024-118, approved April 26, 2024, effective July 1, 2024) enacted the council and the generated child pornography crimes, but not the AI transparency and disclosure rules.
Information Technology Act Amendments
Utah's SB 131 requires paid election communications that contain AI-generated synthetic media to carry set disclosures. Audio must say "Contains content generated by AI." at the beginning and end; visuals must show a set AI label throughout the synthetic portions; online versions must also carry embedded tamper-evident content provenance data. A court may impose a civil penalty of up to $1,000 per violation on the creator or sponsor. Judges and the parole board may treat committing or facilitating a crime with the intentional or knowing use and material assistance of AI as an aggravating factor. Signed March 13, 2024; effective May 1, 2024 (the sentencing provision July 1, 2024).
An act relating to creating oversight and safety standards for developers and deployers of inherently dangerous artificial intelligence systems
Vermont's H.341 (2025-2026 biennium) would have set oversight and safety standards for developers and deployers of inherently dangerous AI systems, meaning high-risk AI systems, dual-use foundational models and generative AI systems, used by non-small businesses doing business in Vermont. Duties included reasonable care, NIST-based testing and risk management, and Safety and Impact Assessments filed with an AI Division in the Agency of Digital Services. Harmed consumers could sue. It was referred to House Commerce on February 25, 2025 and died.
An act relating to artificial intelligence
Vermont's H.821 (2026) would have required anyone using generative AI to interact with consumers to clearly disclose, when asked, that the consumer is dealing with AI, and would have required licensed professionals to disclose high-risk generative AI interactions. Violations would have been unfair and deceptive acts enforced by the Attorney General, and a person could not blame generative AI as a defense. It would also have added members to the Artificial Intelligence Advisory Council and required a report by January 15, 2027. Referred to committee January 29, 2026; died.
Artificial Intelligence Amendments
Utah's HB 452 regulates mental health chatbots: generative AI that holds therapy-like conversations and that the supplier represents, or a reasonable person would believe, can provide mental health therapy or help manage mental health conditions. Suppliers may not sell or share Utah users' identifiable health information or input (limited exceptions), may not use input to target or tailor ads, must label in-chat ads, and must disclose the chatbot is AI before access, after seven days of non-use and when asked. Fines up to $2,500 per violation. Signed March 25, 2025; effective May 7, 2025.
To create an AI Task Force
West Virginia's HB 5490 (2024 regular session) would have created a Generative AI and Natural Language Processing Task Force (new W. Va. Code 5A-6-9) set up by the Governor's Office, to meet at least quarterly and report to the Governor and the Legislature by July 1, 2025. It would recommend consumer-information protections, school policies on student use of generative AI, public-service uses, civil-liberties protections, workforce effects and cybersecurity. It was sent to House Artificial Intelligence on February 6, 2024 and went no further.
Creating the protection against deep fakes act to provide a cause of action for damages arising from the use of generative artificial intelligence to create an image or likeness of another person without such person's consent for use in obscene material.
Not law. Kansas SB 525, the Protection Against Deep Fakes Act, would have let a person whose image or likeness is shown through synthetic media in published obscene material, without express written consent, sue the producer, promoter or intentional possessor. A plaintiff could recover actual damages, punitive damages equal to the defendant's revenue from the material, and attorney fees. It would not be a defense that the likeness was AI-generated and is not the person's actual image or likeness. It was referred to Senate Judiciary February 21, 2024 and died in committee April 30, 2024.
Provides for the regulation of mental health chatbots that use artificial intelligence technology. (gov sig)
Louisiana's SB5 would have required operators of AI mental health chatbots to disclose clearly that the chatbot is AI and not a human, to maintain protocols for suicidal ideation, self-harm and harm to others, and not to sell or share users' identifiable health information. It would also have restricted advertising in chatbot conversations. The Attorney General could seek fines up to $10,000 per violation after a 45-day cure notice. It stayed in Senate Health and Welfare and died.
Prohibits the development of artificial intelligence systems with certain capabilities involving interactions with minors (OR SEE FISC NOTE GF EX)
Louisiana's HB295, the Guidelines for User Age-Verification and Responsible Dialogue Act, would have made it a crime to knowingly or recklessly design, develop or deploy an AI chatbot able to solicit sexual conduct from minors, send them sexual images, or encourage suicide, self-injury or violence (up to 1 year and $100,000). It would also have required age verification, AI-companion bans for minors and recurring AI disclosures. It was referred to House Commerce March 9, 2026 and died.
Establishing protections for consumers interacting with artificial intelligence chatbots
Massachusetts S264 would add G.L. Chapter 93M requiring any commercial entity that deploys a chatbot to clearly and conspicuously disclose that the user is dealing with a chatbot and not a human, and making chatbot statements legally equivalent to those of an employee or agent. Reported favorably and referred to Senate Ways and Means December 18, 2025; it has not passed either chamber.
Use of fabricated media in elections.
Indiana's SB 7 (2024) would have barred disseminating 'fabricated media' made with generative AI, or contracting to do so, fewer than 90 days before an election to injure a candidate or influence the outcome, when the person knows or reasonably should know it is fabricated, the depicted person did not consent (for altered recordings and imitations of real people), and it lacks the disclaimer 'Elements of this media have been digitally altered or artificially generated.' A knowing and intentional violation would be a Class B misdemeanor, with enhancements. Referred to committee January 8, 2024; it did not pass.
Use of altered media in elections.
Indiana's HB 1283 (2024) would have barred anyone who finances a candidate communication from disseminating 'fabricated media' (altered recordings or generative-AI imitations made without consent) of a candidate in an election within 90 days, to injure the candidate or party or influence the outcome, unless it carries the disclaimer 'This media has been manipulated or generated by artificial intelligence.' A depicted candidate could sue; the bill sets no criminal penalties. Referred to committee January 9, 2024; it did not pass.
Eavesdropping, Peeping, and Voyeurism
South Carolina's H. 5394 (2023-24) would have amended Code 16-17-470 (eavesdropping, peeping and voyeurism) to increase penalties and to cover knowingly creating or altering images, audio, video or digital files with a generative adversarial network or other digital technology to make a realistic but false depiction of another person without consent, for sexual arousal or gratification, and knowingly selling or distributing such material without consent. It would also have required destruction of evidence and notice to additional victims. Referred to House Judiciary April 10, 2024; it did not pass.
Relating to the artificial intelligence division within the Department of Information Resources.
Texas SB 1700 (2025), as introduced, would have had the Department of Information Resources establish an artificial intelligence division to help state agencies use generative AI on projects such as modernizing or replacing legacy systems. Projects could use department-developed AI or a vendor under a contract requiring the vendor's AI to complete at least 90 percent of the work, and the division would write cost analysis reports. No enrolled version exists, so it died June 2, 2025.
Relating to prohibiting the use of certain political advertising manipulated by generative artificial intelligence technology; creating a criminal offense.
Texas SB 228 (2025), as introduced, would have added Election Code Sec. 255.009 making it a Class B misdemeanor to cause political advertising to be published, distributed or broadcast when it includes an image, audio or video of an officeholder's or candidate's appearance, speech or conduct intentionally manipulated using generative AI to create a realistic but false depiction. The bill has no pre-election window, disclosure safe harbor, exemptions or definitions. No enrolled version exists, so it died June 2, 2025.
CHAT Act Children Harmed by AI Technology Act
The CHAT Act would require operators of companion AI chatbots (built mainly to simulate friendship, companionship, emotional or therapeutic conversation) to require user accounts and verify every user's age; existing accounts would be frozen until verified. Minors' accounts must be tied to a verified parent account with parental consent, parents must be told immediately of conversations involving suicidal ideation, and minors must be blocked from chatbots that engage in sexually explicit communication. A pop-up must say the user is not talking to a human at the start and every 60 minutes. Enforced by the FTC and state attorneys general; effective 1 year after enactment.
Enacting the Kansas age-appropriate design code act to require businesses to assess and mitigate risks of compulsive use in minors; enacting the Kansas stopping likeness abuse by nonconsensual digital replicas act to create a private right of action for the unauthorized digital replication and…
Not law. HB 2772 would have enacted three Kansas acts. An Age-Appropriate Design Code Act would have required online businesses likely to be used by minors to assess designs for the risk of compulsive use, use the lowest-risk design by default, give minors the highest privacy settings, and limit data collection. A digital-replica act would have let people sue over digital replicas of their likeness made without consent. A chatbot act would have required chatbots to say they are not human, limited the use of chat logs, and made providers liable for injuries their chatbots cause. It died in the House Federal and State Affairs Committee on April 10, 2026.
Enacting the Kansas age-appropriate design code act to require businesses to assess and mitigate risks of compulsive use in minors, enacting the Kansas stopping likeness abuse by nonconsensual digital replicas act to create a private right of action for the unauthorized digital replication and…
Not law. SB 499, the Senate twin of HB 2772, would have enacted three Kansas acts. An Age-Appropriate Design Code Act would have required online businesses likely to be used by minors to assess designs for the risk of compulsive use, use the lowest-risk design by default, give minors the highest privacy settings, and limit data collection. A digital-replica act would have let people sue over replicas of their likeness made without consent. A chatbot act would have required chatbots to say they are not human, limited the use of chat logs, and made providers liable for injuries their chatbots cause. It died in the Senate Federal and State Affairs Committee on April 10, 2026.
Establishing the Kansas community harmed by AI technology act, mandating user accounts and age verification for AI chatbot access, classifying users by age, requiring parental consent for minors, blocking explicit content, protecting age information confidentiality, monitoring for suicidal…
Not law. The Kansas CHAT Act (Community Harmed by AI Technology Act) would have required anyone offering a companion AI chatbot in Kansas to require user accounts, verify every user's age (freezing existing accounts until verified), and link minors' accounts to a verified parent account with parental consent. Minors would have been blocked from chatbots that engage in sexually explicit talk, and cut off with their parent alerted if they expressed suicidal thoughts. Chatbots would have had to show a pop-up at least every 60 minutes saying the user is not talking to a human. It died in committee on April 10, 2026.
Making it unlawful for a person to knowingly train artificial intelligence to encourage or support suicide or the unlawful killing of another person, provide emotional support, develop emotional relationships, act as a healthcare professional, simulate humans or encourage isolation.
Not law. SB 405 would have made it unlawful in Kansas to knowingly train AI, including chatbots, to encourage suicide or killing, give emotional support, act as a companion, pose as a licensed mental health or healthcare professional, act like a human friend, encourage isolation from family, or simulate a human being. The Attorney General or a harmed person could have sued for actual damages or $150,000, plus punitive damages and fees; courts could order the AI shut down until retrained and fine up to $50,000 per violation. It had a hearing on February 9, 2026 and died in committee on April 10, 2026.
Creates provisions relating to artificial intelligence chatbots
Not law. SB 1455, Missouri's Senate version of the GUARD Act, would have made it unlawful to design or offer an AI chatbot knowing or recklessly disregarding that it risks soliciting minors into sexually explicit conduct, or that it encourages suicide, self-injury or violence (fines up to $100,000 per offense). Operators would have had to require accounts, verify every user's age, freeze existing accounts until verified, bar minors from their chatbots, and disclose at the start of each conversation and every 30 minutes that the chatbot is AI, not a human. It was referred to committee and got no further action before the May 15, 2026 deadline for passing bills.
Creates a consumer bill of rights regarding artificial intelligence
Not law. HB 734 would have created a Louisiana AI bill of rights. Companion chatbot platforms would have needed a parent's consent for minors' accounts, with parental controls and self-harm alerts, AI disclosures and hourly break reminders for minors, and safeguards against material harmful to minors. Any bot would have had to show an hourly pop-up saying the user is not talking to a human. It also would have barred AI companies from selling personal data unless deidentified, banned commercial use of AI-generated likenesses without consent, and limited state AI contracts with entities tied to foreign countries of concern. It was withdrawn on March 30, 2026.
Regulates artificial intelligence in news media industry; establishes "Artificial Intelligence In Communications Oversight Committee."
New Jersey's A3307 would regulate AI in the news media. News entities could use AI to assist staff but not replace them, and generative AI content would need a prominent label, source credit and an accuracy disclaimer. Civil penalties would be $10,000, $20,000 and then $30,000 per violation. It would also create the Artificial Intelligence In Communications Oversight Committee. Pre-filed for introduction in the 2026 session.
AN ACT to amend Tennessee Code Annotated, Title 29; Title 37 and Title 47, relative to artificial intelligence.
Tennessee Public Chapter 1082 directs the Tennessee Advisory Commission on Intergovernmental Relations (TACIR) to study how the state might regulate AI systems and generative AI chatbots. The study must cover federal law, other states' approaches, constitutional limits, costs to government, economic effects on businesses, developers, schools and consumers, and safeguards for minors and mental health, including suicide prevention, with any recommendations for legislation. Introduced as the Curbing Harmful AI Technology (CHAT) Act regulating companion chatbots, it was rewritten into this study. Signed May 22, 2026; effective immediately.
AN ACT to amend Tennessee Code Annotated, Title 29; Title 37 and Title 47, relative to artificial intelligence.
HB 1946 is the House companion of SB 1700, which became Tennessee Public Chapter 1082 after the House passed the Senate bill in its place on April 21, 2026. The law directs the Tennessee Advisory Commission on Intergovernmental Relations (TACIR) to study how the state might regulate AI systems and generative AI chatbots: federal law, other states' approaches, constitutional limits, costs, economic effects, and safeguards for minors and mental health, including suicide prevention, with any recommendations for legislation. It began as the Curbing Harmful AI Technology (CHAT) Act. Signed May 22, 2026; effective immediately.
Artificial intelligence systems that simulate humanlike relationships with children and providing a penalty.
Not law. AB 965 would have barred operators from making a companion chatbot available to Wisconsin children unless safety measures kept it from encouraging self-harm, suicide, violence, drug or alcohol use or disordered eating; offering mental health services; encouraging illegal acts; sexual content with a child; putting validation ahead of accuracy or safety; or optimizing engagement over those safeguards. Until 2027 it would have applied only when the operator knew the user was a child. Forfeitures up to $25,000 per violation, and harmed children could sue. It failed to pass when the session's floor period ended on March 23, 2026.
A bill for an act relating to the requirements for chatbot deployers, including required protocols, limitations on data collection, and requirements for minors to interact with artificial intelligence companions and therapeutic chatbots, and providing civil penalties, punitive penalties, and civil…
Not law. HF 2204 would have required anyone offering a chatbot to the public in Iowa to keep protocols to detect, respond to and reduce harm to users, putting user safety first, and to collect only the data it needs. Deployers would have had to use age verification to keep minors from AI companions (chatbots that simulate a romantic or emotional bond), get permission before offering chatbots that impersonate real people, and meet strict conditions, including a licensed professional's recommendation and clinical trial evidence, before offering therapeutic chatbots to minors. It was referred to committee on January 29, 2026 and died when the 2026 session adjourned.
Law Enforcement Artificial Intelligence Amendments
Utah's SB 205 (2026), Law Enforcement Artificial Intelligence Amendments (second substitute), would have amended the 2025 law on police use of generative AI to require agency policies to cover available oversight or safeguard settings and to require agencies to post their policy publicly. It passed the Senate on February 26, 2026 (20-0-9); the House struck the enacting clause on March 6 and the Senate filed it.
Requires advertisements to disclose the use of a synthetic performer
New York requires advertisers to disclose when an ad contains a synthetic performer: a digital asset created or modified with generative AI or software to look like a human performer who is not any identifiable real person. Anyone who produces or creates an ad for goods or services, in any medium, must conspicuously disclose it if they actually know the synthetic performer is there. Penalties are $1,000 for a first violation and $5,000 for each later one. Audio ads, AI used only to translate a performer, and promos for films, shows or games that use the performer as the work does are exempt, as are the outlets that run ads. Signed December 11, 2025 (Chapter 617); effective June 9, 2026.
Artificial Intelligence Fraud and Accountability Act; enact.
Mississippi's SB 2354 (2026), the "Artificial Intelligence Fraud and Accountability Act," would have created a civil cause of action for anyone injured by use of AI systems (including deepfakes, voice cloning, generative tools, chatbots or automated decision-making systems) to deceive, impersonate or obtain money, property, data or other assets. It would have allowed the greater of compensatory damages or $500 per violation. It was referred to Judiciary, Division A on January 19, 2026, and died in committee on February 3, 2026.
Mental health professionals: artificial intelligence.
Vetoed. SB 903, the Wellness and Oversight for Psychological Resources Act, would have limited AI in California psychotherapy to administrative and supplementary support. Recording or transcribing sessions, or AI triage, would have needed the patient's informed consent, and AI could not make therapeutic decisions, produce diagnoses or treatment plans, detect emotions or screen patients without a licensed professional's review. Companion chatbots could not be marketed as therapy. Governor Newsom vetoed it on September 30, 2026, calling it overly broad and saying it would restrict useful clinical tools and reach general-purpose AI.
Toys: companion chatbots.
California bans toys with companion chatbots until 2031. From January 1, 2027, no one may make, sell, exchange, possess for sale, or offer to a retailer a toy (a physical product designed, marketed or made for play by children under 16) that includes a companion chatbot: an AI system with a natural-language interface that gives adaptive, human-like responses and can meet a user's social needs, including by sustaining a relationship across interactions. Violations carry the same civil liability as the state's companion chatbot rules. The ban is repealed January 1, 2031. Signed September 10, 2026 (Chapter 189).
Adds to existing law to establish provisions regarding unbiased artificial intelligence in state government purchasing.
Did not pass. H 687 would have let Idaho executive-branch agencies, including public colleges, buy or use a large language model only if it is 'truth-seeking' and 'ideologically neutral.' Agencies could not use models designed to alter facts or history to advance diversity, equity and inclusion (DEI) outcomes, restrict lawful content based on race or sex, embed DEI-based guardrails, or require users to affirm DEI concepts. Contracts would have given agencies access to system prompts, training directives and test results to check compliance. It passed the House 56-11 but died in the Senate State Affairs Committee.
Artificial intelligence disclosure statements.
North Dakota requires a disclaimer on political content made with AI to impersonate a person. Any communication or advertising for a political purpose (supporting or opposing a candidate, political committee or party, or an initiated or referred measure or petition) that contains images, graphics, video, audio, text or other digital content created in whole or in part with AI to visually or audibly impersonate a human must prominently state: 'THIS CONTENT GENERATED BY ARTIFICIAL INTELLIGENCE.' AI used only for text generation, grammar, spelling or editing that creates no new impersonation is exempt. Signed April 10, 2025; effective August 1, 2025.
Companion chatbots: children’s safety.
California's SB 1119, "Adam's Law," sets child-safety rules for companion chatbots. Operators must determine users' ages under the Digital Age Assurance Act or give every user the child protections. From July 1, 2027, operators that allow children must assess risks before releasing new or substantially modified chatbots, publish a child safety policy, and provide crisis protocols, parent-only default settings, parental controls, AI notices and limits on ads and data use. Independent audits follow. Civil penalties reach $15,000 per affected child. Signed September 10, 2026 (Chapter 190); effective January 1, 2027, with most duties operative July 1, 2027.
Leading Ethical AI Development (LEAD) for Kids Act.
Vetoed. California's AB 1064, the LEAD for Kids Act, would have barred businesses and government agencies from making a companion chatbot available to a child unless it was not foreseeably capable of encouraging self-harm, suicidal ideation, violence, drug or alcohol use or disordered eating; offering unsupervised therapy; encouraging harm or illegal acts; sexually explicit interactions; favoring validation over accuracy or safety; or optimizing engagement over those guardrails. The Attorney General could seek $25,000 per violation and harmed children or parents could sue. Vetoed by Governor Newsom October 13, 2025; veto consideration stricken January 22, 2026.
Companion chatbots.
California's SB 243 regulates companion chatbots, AI systems that give adaptive, human-like responses and can meet a user's social needs. Operators must disclose that the chatbot is AI if a reasonable person would be misled to think it is human, keep and publish a protocol against suicide and self-harm content with crisis referrals, and warn that companion chatbots may not suit some minors. Known minors get AI disclosures, break reminders at least every three hours and sexual-content safeguards (deleted January 1, 2027 by SB 1119). Annual reports to the Office of Suicide Prevention start July 1, 2027. Injured persons may sue. Signed October 13, 2025 (Chapter 677); effective January 1, 2026.
Artificial intelligence models: large developers.
California's SB 53, the Transparency in Frontier Artificial Intelligence Act, covers developers of frontier models (foundation models trained with more than 10^26 operations). Large frontier developers (over $500 million in prior-year revenue with affiliates) must publish and follow a frontier AI framework for catastrophic risks. All frontier developers must publish transparency reports at deployment, report critical safety incidents to the Office of Emergency Services within 15 days, and not retaliate against whistleblowers. Large developers face penalties up to $1 million per violation. Signed September 29, 2025 (Chapter 138); effective January 1, 2026.
Generative Artificial Intelligence Accountability Act.
California's SB 896, the Generative Artificial Intelligence Accountability Act, governs state government's use of generative AI. The Department of Technology updates the state's GenAI report under Executive Order N-12-23 as needed; the Office of Emergency Services, as appropriate, analyzes GenAI threats to critical infrastructure, including mass-casualty events, with an annual summary to the Legislature. Agencies using GenAI to communicate directly with people about government services and benefits must disclose it and say how to reach a human. Signed September 29, 2024 (Chapter 928); effective January 1, 2025.
USE OF AI BY STATE GOVERNMENT
Illinois' SB2117 would amend the Department of Innovation and Technology Act to expand the Generative AI and Natural Language Processing Task Force so that it also recommends policies for State agencies' procurement, deployment and use of generative AI, recommends prohibited agency uses, and assesses the effect of such use on State services. Filed February 7, 2025 and referred to the Senate Assignments Committee; it has not passed either house.
An Act To Amend Titles 10 And 11 Of The Delaware Code Relating To Deep Fakes.
Delaware's HB 353 (84 Del. Laws c. 479), the Amelia Kramer Act, extends Delaware's intimate-image laws to deep fakes, meaning realistic but false sexual images made with generative adversarial network techniques or other digital technology. An identifiable person harmed by the intentional disclosure or threatened disclosure of a deep fake without consent may sue for damages or statutory damages up to $10,000, the defendant's gains, punitive damages, fees and an injunction. The violation-of-privacy crime now covers deep fakes without proof of an expectation of privacy, and an adult creating a depiction of a minor is an aggravating factor. Signed and effective October 9, 2024.
Companion chatbots.
California's SB 300 (2025-26 Regular Session) would amend Business and Professions Code section 22602 on companion chatbots. It would require a companion chatbot operator to take its existing protective actions for minors when it has constructive knowledge that a user is a minor, and to prevent the chatbot from producing or facilitating the exchange of any sexually explicit material or proposing sexually explicit conduct. Passed the Senate 38-0; placed on the Assembly inactive file August 30, 2026; not passed.
SCH CD-CYBER-BULLYING
Illinois' Public Act 104-0338 (HB 3851) amends the School Code's bullying prevention section (Section 27-23.7). Beginning with the 2026-2027 school year, "cyber-bullying" also includes posting or distributing an unauthorized digital replica, meaning a digital replica of an individual used without the depicted individual's consent, by electronic means if it causes any of the harms in the bullying definition. "Digital replica" has its meaning in the Digital Voice and Likeness Protection Act, which covers replicas made with AI or other technology. The act also lists posting or distributing sexually explicit images as a form of bullying. Approved August 15, 2025; effective July 1, 2026.
LIMIT PREDICTIVE ANALYTICS USE
Illinois' Public Act 103-0804 (HB 3773) amends the Illinois Human Rights Act. It is a civil rights violation for an employer to use artificial intelligence that has the effect of subjecting employees to discrimination on the basis of protected classes in recruitment, hiring, promotion, renewal of employment, selection for training or apprenticeship, discharge, discipline, tenure, or the terms, privileges or conditions of employment, or to use zip codes as a proxy for protected classes. Failing to notify employees that the employer uses AI for these purposes is also a violation; the Department of Human Rights must adopt rules on notice. Approved August 9, 2024; effective January 1, 2026.
AI Chatbots - Licensing/Safety/Privacy
North Carolina's S 624 (as filed) would require a Department of Justice license to operate or distribute a chatbot that deals substantially with health information, with a $50,000 civil penalty for violating G.S. 114B-5. It would also create a Chatbot Safety and Privacy Act: covered platforms would owe users a duty of loyalty and would have to disclose that a chatbot is not human, get consent each session and de-identify user data. Filed March 25, 2025 and referred to the Senate Rules and Operations Committee; it has not passed either house.
ARTIFICIAL INTELLIGENCE SAFETY
Illinois's SB 3261, the Artificial Intelligence Public Safety and Child Protection Transparency Act, would require large frontier developers (models trained with over 10^26 operations; $500 million revenue) and large chatbot providers ($25 million revenue; 1,000,000+ monthly users, likely accessed by minors) to publish safety plans, and frontier developers and large chatbot providers to report safety incidents to the Attorney General. Civil penalties up to $1,000,000 (large developers) or $50,000 (chatbot providers) per violation. Introduced February 3, 2026; re-referred to Assignments May 22, 2026 under Rule 3-9(a); not passed. Would take effect January 1, 2027.
Providing for consumer protection and for artificial intelligence and chatbots; imposing duties on the Bureau of Consumer Protection in the Office of Attorney General; and imposing penalties.
Pennsylvania's HB 2175 (Printer's No. 2831) would add Chapter 71 to Title 12 on AI chatbots that use generative AI to converse with consumers and are represented as helping manage a situation or treat a condition, including mental health. Suppliers could not sell or share health information or consumer input except as allowed, could not advertise in chatbot conversations, and would have to adopt, file and follow a written disclosure policy. Fines up to $2,500 per violation. Introduced January 30, 2026; re-referred to Consumer Protection, Technology & Utilities March 19, 2026; not passed.
Prohibits advertising generative artificial intelligence as able to practice regulated profession or occupation.
New Jersey's A 4733 (as introduced) would bar a person or entity that develops or deploys generative AI in the State from advertising or representing to the public that the AI can practice a profession or occupation regulated under Title 45 of the Revised Statutes or by the Judicial Branch or Administrative Office of the Courts. A violation would be an unlawful practice under the consumer fraud act. Introduced March 10, 2026; reported by two Assembly committees and referred to Assembly Appropriations May 28, 2026; not passed by either house.
"GAI Accountability Act;" imposes civil penalties on generative artificial intelligence platforms engaging in harmful activity, including exploitation of children.
New Jersey's A 5090, the GAI Accountability Act, would make owners, operators and developers of generative AI platforms that do business in, are based in, or are accessible to residents of New Jersey civilly liable if they fail to prevent the platform from engaging in conduct that would be murder, assault, theft by deception or extortion, endangering the welfare of children, or creating child sexual abuse or exploitation material if done by a human. The penalty would be $20,000 per violation. Introduced May 14, 2026 and referred to the Assembly Science, Innovation and Technology Committee; it has not passed either house.
"AI Image Disclosure Act"; concerns disclosure of certain AI-generated content.
New Jersey's A 5089, the AI Image Disclosure Act (as introduced), would require a covered provider of a publicly accessible generative AI system to put a manifest (clear and conspicuous) and a latent (not visible or audible) disclosure in AI-generated images, video and audio. Social media platforms with at least five million account holders worldwide would have to label detected provenance data and could not strip it. Civil penalties would be $5,000 per violation. Introduced May 14, 2026 and referred to the Assembly Science, Innovation and Technology Committee; it has not passed either house.
Safety - As introduced, enacts the "Artificial Intelligence Public Safety and Child Protection Transparency Act." - Amends TCA Title 4; Title 10, Chapter 7; Title 47; Title 58 and Title 68.
Tennessee's HB 1898 (114th General Assembly), the "Artificial Intelligence Public Safety and Child Protection Transparency Act," as introduced would have barred a frontier model developer or a large chatbot provider (1 million monthly users, $25 million revenue) from making materially false or misleading statements about catastrophic risk, and defined child safety incidents involving covered chatbots. The House passed it as amended, 94-0, on April 16, 2026, and the Senate received it April 20, 2026, but it did not pass the Senate before the session ended. Companion: SB 2171.
Prohibits state agencies and state-owned entities from utilizing large language models or any artificial intelligence systems to make decisions that affect individuals' rights, benefits, or services; requires such decisions to be made by human personnel of the respective state agency or state-owned…
New York's A 7278 would add State Technology Law § 106-c, barring state agencies and state-owned entities from using large language models or any AI system to make decisions that affect individuals' rights, benefits or services; those decisions would have to be made by human personnel. Purely advisory use is allowed if a person makes the final decision, and AI may be used for data analysis and research. The attorney general could investigate and enforce. Introduced March 21, 2025 and referred to the Science and Technology Committee (again January 7, 2026); it has not passed either chamber.
Chatbot regulations; personal data; requirements
Arizona's HB 2737 (2026) would have limited how a chatbot provider may use personal data and chat logs: no processing personal data to inform output without a user's express request and affirmative consent, no using chat logs for advertising, no training on chat logs of minors without parental consent or of adults without consent, no selling chat logs, and no retention beyond ten years unless required. It was held in House committees with no action and died when the 2026 session ended June 13, 2026.
Regulating artificial intelligence training data.
Washington's HB 2503 (substitute, 2026 session) would have required developers of generative AI systems to post a high-level summary of their training datasets on their website on or before January 1, 2027 and before each later release or substantial modification made available to Washingtonians. It covered systems released on or after January 1, 2022, and treated violations as Consumer Protection Act violations. It was reported from House Technology, Economic Development, & Veterans on January 28, 2026, referred to Appropriations January 30 and died at adjournment.
HEALTH AI Act Healthcare Enhancement And Learning Through Harnessing Artificial Intelligence Act
The federal H.R. 5045 (HEALTH AI Act) would require the Secretary of Health and Human Services to establish a grant program for research on the use of generative AI in health care. Grants would go to universities, 501(c)(3) nonprofits and federal, State, local or Tribal agencies, with priority for entities that encourage adoption, invest in workforce development, reduce burnout or health disparities, or serve underserved populations. Introduced August 26, 2025 and referred to the House Energy and Commerce Committee; no further action.
CHAT Act (Children Harmed by AI Technology Act)
The CHAT Act would require anyone offering a companion AI chatbot in the U.S. to require user accounts, verify every user's age (freezing existing accounts until verified), and link minors' accounts to a verified parent account with parental consent. Parents would be told immediately of any conversation involving suicidal ideation, and minors blocked from chatbots that engage in sexually explicit talk. Chatbots would show a pop-up at the start of each conversation and at least every 60 minutes saying the user is not talking to a human. The FTC and state attorneys general would enforce it. Pending in the Senate Commerce Committee since September 4, 2025.
DOIT-AI TASK FORCE
Illinois's HB 3646 (104th General Assembly) would amend the Department of Innovation and Technology Act's Generative AI and Natural Language Processing Task Force. The engrossed bill would require at least 5 hybrid public meetings, one each in Chicago, Springfield, the Metro East, the Quad Cities and Southern Illinois. Passed the House 111-0 on April 11, 2025; re-referred to Senate Assignments June 2, 2025.
General Regulatory Provisions -- Generative Artificial Intelligence Models
Rhode Island H 6286 (2023), as introduced on April 19, 2023, would add chapter 6-59 setting operating standards for companies running large-scale generative AI models (at least one billion parameters). Companies would have to watermark or authenticate text output, obtain informed consent before using personal data, run risk assessments and register with the attorney general within 90 days. The text sets no penalties and takes effect September 1, 2023.
Education-Tech
Illinois's HB 299 (103rd General Assembly) would have amended the School Code's bullying prevention provisions so that bullying includes posting or distributing sexually explicit images and, beginning with the 2025-2026 school year, cyber-bullying includes distributing or posting a digital replica of an individual. It passed the House 112-0-0 on May 22, 2024, was referred to Senate Assignments, and was not passed before the session ended (Session Sine Die, January 7, 2025).
PUBLICITY ACT-USE OF AI
Illinois's SB 3325 (103rd General Assembly) would have amended the Right of Publicity Act to give recording artists additional enforcement rights and remedies, make liable anyone who materially contributes to, induces or facilitates a violation after having reason to know of it, and define AI and generative AI. It passed the Senate Judiciary Committee 7-0 but died at Session Sine Die on January 7, 2025.
HEALTH CARE GENERATIVE AI USE
Illinois's SB 2259 (104th General Assembly) would add Section 67 to the Medical Practice Act of 1987. A health facility, clinic, physician's office or group practice office that uses generative AI to produce written or verbal patient communications about clinical information would have to ensure the communications meet specified criteria; communications read and reviewed by a human licensed or certified provider would be exempt. Re-referred to Assignments April 11, 2025.
CHATBOT PROVIDER-LIABILITY
Illinois's HB 5044 (104th General Assembly) would create the Chatbot Provider Liability Act. Chatbots would be deemed products for strict liability, and a chatbot provider would have a duty to ensure its chatbot does not injure a user and would be liable for injuries it causes a user even if it used all reasonable care or did not distribute the chatbot directly. Harmed users could sue for damages, injunctive relief and attorney's fees. Re-referred to Rules March 27, 2026.
PROVENANCE DATA REQUIREMENTS
Illinois's HB 4711 (104th General Assembly) would create the Provenance Data Requirements Act. A generative AI tool provider would have to apply provenance data, directly or through third-party technology, to wholly-generated synthetic content its tool produces, and the bill would set additional requirements for large online platforms (1,000,000 or more unique monthly users) and capture device manufacturers. Re-referred to the House Rules Committee March 27, 2026.
FRAUD-ARTIFICIAL INTELLIGENCE
Illinois's SB 1792 (104th General Assembly) would add Section 2HHHH to the Consumer Fraud and Deceptive Business Practices Act. The owner, licensee, or operator of a generative AI system would have to conspicuously display a warning on the system's user interface, reasonably calculated to consistently apprise users that outputs may be inaccurate or inappropriate. A violation would be an unlawful practice. Last action: re-referred to Assignments April 11, 2025.
PROVENANCE DATA REQUIREMENTS
Illinois's SB 1929 (104th General Assembly) would create the Provenance Data Requirements Act. A generative AI tool provider would have to apply provenance data, directly or through third-party technology, to wholly generated synthetic content from its tool, and the bill would set further requirements for generative AI tool providers, large online platforms and capture device manufacturers. Last action: re-referred to Assignments April 11, 2025.
Companion chatbots: children’s safety.
California's AB 2023 (2025-26 Regular Session) would have required an operator of a companion chatbot to, on or before July 1, 2027, take child-safety steps including an annual risk assessment, and to submit to an independent audit whose report goes to the Attorney General. The Assembly passed it 66-8 on May 26, 2026; it was ordered to the Senate inactive file on August 30, 2026 and was not passed before the session ended.
Digital Dignity Act.
California's SB 1142 (2025-26 Regular Session), the Digital Dignity Act, would have imposed specified liability on a person who distributes content with actual knowledge it uses a digital replica and is found guilty of a crime requiring false impersonation or liable for defamation arising from the replica. It would also have required large online platforms with a generative AI tool to let individuals report digital replica content and to remove it on a court order. It was held in Senate Appropriations on May 14, 2026 and was not passed before the session ended.
Income tax: credits: commercial production.
California's AB 2403 (2025-26 Regular Session) would have created a 20% or 30% income tax credit for qualified production costs above $500,000 for commercials produced in California, for taxable years from January 1, 2027 to before January 1, 2032. It would exclude commercials created entirely by generative AI or using generative AI or autonomous vehicles in ways that replace human job functions. Held in Senate Appropriations on August 13, 2026; not passed before the session ended.
Department of Technology: Office of Artificial Intelligence: state agency public interface: use of AI.
California's SB 313 (2023-24 Regular Session), the California AI-ware Act, would create an Office of Artificial Intelligence in the Department of Technology to guide state agencies' design, use and deployment of automated systems consistent with privacy and civil liberties law and to minimize bias. A state agency using generative AI to directly communicate with a person would have to give notice and instructions for reaching a human. Held in Senate Appropriations May 18, 2023; not passed.
State agencies or departments: public communications.
California's AB 2412 (2025-26 Regular Session) would amend Government Code section 11549.66 to require a state agency or department that uses generative AI to directly communicate with the public, whether with a specific member of the public or in a general public announcement, to include a GenAI disclaimer and information on how to contact a human employee. Passed the Assembly 73-0; its final Senate committee hearing was canceled at the author's request June 26, 2026; not passed.
Use of artificial intelligence-based tools; covered artificial intelligence, disclosure of use.
Virginia's HB 1294 (2026 Regular Session) would amend Code 19.2-11.14 to require police reports to disclose any use of covered artificial intelligence in a criminal investigation, and to require that AI-assisted report drafts be retained. Continued to 2027 in the House Communications, Technology and Innovation Committee on February 2, 2026.
Law-enforcement agencies; use of certain technology & interrogation practices; forensic laboratory.
Virginia's HB 1261 (2026 Regular Session) would direct the Department of Criminal Justice Services to set a framework for law-enforcement use of generative AI, machine learning systems and audiovisual surveillance technologies and for interrogations, with model policies and training. Continued to 2027 in committee on February 2, 2026.
Artificial Intelligence Training Data Transparency Act; transparency and disclosure requirements.
Virginia's HB 2250 (2025 Regular Session) would have amended the Consumer Data Protection Act and added an Artificial Intelligence Training Data Transparency Act (new Title 59.1 chapter, §§ 59.1-607 through 59.1-611) with requirements for generative AI developers. It would also have allowed third-party opt-out of data processing and required consent to process adolescents' data. Left in committee on February 4, 2025; it failed.
Law-enforcement agencies; use of certain technologies and interrogation practices.
Virginia's HB 2433 (2025 Regular Session) would have had the Department of Criminal Justice Services set a framework for law-enforcement use of generative AI, machine learning systems, audiovisual surveillance technologies and interrogations, with model policies by January 1, 2026 and agency policies by July 1, 2026. It also covered forensic laboratory accreditation. Left in committee on February 5, 2025; it failed.
Artificial Intelligence Transparency Act; established.
Virginia's SB 1161 (2025 Regular Session) would have created the Artificial Intelligence Transparency Act, requiring developers of generative AI systems that produce audio, images, text or video to include a clear and conspicuous disclosure on AI-generated content and to prevent downstream use without it. Passed by indefinitely in committee on January 24, 2025; it failed.
Safety - As introduced, enacts the "Artificial Intelligence Public Safety and Child Protection Transparency Act." - Amends TCA Title 4; Title 10, Chapter 7; Title 47; Title 58 and Title 68.
Tennessee's SB 2171 (114th General Assembly), the "Artificial Intelligence Public Safety and Child Protection Transparency Act," as introduced would have barred a frontier model developer or a large chatbot provider (1 million monthly users, $25 million revenue) from making materially false or misleading statements about catastrophic risk, and defined child safety incidents involving covered chatbots. Companion HB 1898 passed the House. SB 2171 was not passed before the session ended in April 2026; last action: Re-refer to Senate Commerce & Labor Committee (April 15, 2026).
Commerce and trade; require that AI generated content include a disclaimer indicating such content was generated using artificial intelligence
Georgia's HB 478 would add Code section 10-1-393.20 to the Fair Business Practices Act. On or after July 1, 2025, any individual disseminating AI-generated video, audio or still images to facilitate commerce or trade would have to ensure the content discloses that artificial intelligence made all or part of it, in a specified readable or audible form, and omitting the disclosure would be unlawful, each item being a separate violation. The bill reached House Second Readers on February 20, 2025 and died when the session ended.
Fiscal Code - omnibus amendments (Act 45 of 2025), including section 1755-E on stenography services and transcripts created by artificial intelligence
Pennsylvania's Act 45 of 2025 is a Fiscal Code omnibus; its AI provision is new section 1755-E. Money appropriated for stenography services for administrative proceedings may not be used to create a stenographic transcript that is solely and exclusively created by artificial intelligence without a human stenographer present, in person or virtually. Administrative proceedings are actions, proceedings or appeals before an executive branch agency, and AI includes generative AI. The section took effect immediately on approval (Nov. 12, 2025).
Health care professions: deceptive terms or letters: artificial intelligence.
California's AB 489 (Bus. & Prof. Code §§ 4999.8-4999.9) makes the laws barring use of terms, letters or phrases that imply a health care license one does not hold enforceable against anyone who develops or deploys AI or generative AI that uses them in its advertising or functionality. AI technology may not use terms implying that its care, advice, reports or assessments come from a licensed natural person. Each use is a separate violation, under the jurisdiction of the relevant licensing board. Signed October 11, 2025 (Chapter 615); effective January 1, 2026.
Regarding The Ownership Of Model Training And Content Generated By A Generative Artificial Intelligence Tool.
Arkansas Act 927 of 2025 sets who owns what a generative AI tool produces. The person who provides the input owns the generated content, as long as it doesn't infringe existing copyrights or other intellectual property, and owns a model trained with the tool if the training data was lawfully acquired and ownership hasn't been transferred by contract. When an employee uses the tool within the scope of employment and under the employer's direction and control, the employer owns the result. Approved April 21, 2025; effective August 5, 2025.
Disclosures regarding content generated by artificial intelligence in political advertisements, granting rule-making authority, and providing a penalty.
2023 Wisconsin Act 123 requires AI disclosures in political communications. Audio or video containing express advocacy, issue advocacy or support for or opposition to a referendum that contains synthetic media (audio or video substantially produced by generative AI) must say so: audio states "Contains content generated by AI" at the beginning and end, and video displays a disclosure such as "This video content generated by AI." Violations carry a forfeiture of up to $1,000 each. It took effect the day after publication (published March 22, 2024).
2026 Appropriations Act.
North Carolina's 2026 Appropriations Act (S.L. 2026-41) includes Section 7.39, Safe and Responsible AI in Schools. The State Board must adopt K-12 AI literacy standards (responsible use, limits of AI, verifying outputs, privacy, safety with chatbots). DPI must write a model AI policy by Dec. 31, 2026, and local boards, charter, lab and other public schools must adopt AI use policies by June 30, 2027. DPI must keep an evaluation framework and public lists of AI tools, and all teachers must complete AI training by June 30, 2028. The act took effect July 1, 2026.
AN ACT relating to mental health chatbots.
Kentucky's HB 641 (the Artificial Intelligence in Mental Health Act, as introduced) would add sections to KRS Chapter 367 regulating suppliers of mental health chatbots, AI systems that use generative AI to hold therapy-like conversations. Suppliers could not sell or share users' health information or input, would have to label and disclose any advertising, and would have to disclose every 30 minutes that the chatbot is AI. The Attorney General could fine up to $2,500 per violation. Introduced February 13, 2026 and referred to the House Small Business & Information Technology Committee; it died in committee when the session ended.
Requires synthetic content creations system providers to include provenance data on certain content made available by the provider
Passed both houses; awaiting delivery to the Governor. S 6954-B, the Stop Deepfakes Act, would require providers of publicly available generative AI systems to attach provenance data (content credentials such as the C2PA standard) to AI-generated or AI-modified images, audio and video, identifying it as synthetic, the provider and when it was made. Hosting platforms could not knowingly offer systems that don't. Large online platforms would have to keep provenance data and show it to users, and state agencies would label their own media. The Attorney General could seek injunctions and fines up to $25,000 per violation. Effective 180 days after signing.
Establishes the artificial intelligence training data transparency act requiring developers of generative artificial intelligence models or services to post on the developer's website information regarding the data used by the developer to train the generative artificial intelligence model or…
New York's A 6578-B (the "artificial intelligence training data transparency act") would add General Business Law Art. 44-C, §§ 1430-1432. On or before January 1, 2027, and before each later release, a developer of a generative AI model or service (or substantial modification) released on or after January 1, 2022 and available to New Yorkers would have to post a high-level summary of its training datasets on its website. Would take effect immediately. Passed the Assembly and, on June 4, 2026, the Senate (substituted for S6955-A); the page shows no delivery to the Governor.
Requires the owner, licensee or operator of a generative artificial intelligence system to conspicuously display a notice on the system's user interface that is reasonably calculated to consistently apprise the user that the outputs of the generative artificial intelligence system may be inaccurate.
New York's A 3411-B would add General Business Law § 399-zzzzzz requiring the owner, licensee or operator of a generative artificial intelligence system to clearly and conspicuously display a notice on the system's user interface that its outputs may be inaccurate. Failure carries a civil penalty of up to $1,000 for each violation, with each user not given the notice a separate violation. Would take effect on the ninetieth day after it becomes law. Passed the Assembly January 28, 2026 and the Senate March 9, 2026 (58-2, substituted for S934A); no delivery to or action by the Governor appears in the record.
Requires publishers of books created wholly or partially with the use of generative artificial intelligence to disclose such use of generative artificial intelligence before the completion of such sale; applies to all printed and digital books consisting of text, pictures, audio, puzzles, games or…
New York's A 1509 would add General Business Law § 338 requiring any book wholly or partially created through generative artificial intelligence and published in New York to conspicuously disclose on its cover that it was created with the use of generative AI. It covers all printed and digital books, regardless of target audience, made of text, pictures, audio, puzzles, games or any combination. Would take effect on the 60th day after becoming law. Referred to Consumer Affairs and Protection January 7, 2026; it has not passed either chamber.
Attorneys: court filings: artificial intelligence.
California's AB 2811 (2023-24 Regular Session) would add Business and Professions Code section 6068.1, requiring an attorney to execute and keep for seven years an affidavit certifying whether a generative AI program was used in drafting each document filed, or intended to be filed, in a state or federal court in California, and to file the affidavit with a court on request. It was never heard: its April 1, 2024 committee hearing was canceled at the author's request, and it was not passed.
Generative artificial intelligence: training data transparency.
California's AB 2013 (Civil Code §§ 3110-3111) requires developers of generative AI systems or services made publicly available to Californians, free or paid, to post documentation of their training data on their websites: by January 1, 2026 for systems released since January 1, 2022, and before each later release or substantial modification. The documentation summarizes the datasets, including sources, size, data types, and whether they contain copyrighted material or personal information or were bought or licensed. Signed September 28, 2024 (Chapter 817); took effect January 1, 2025, with first postings due January 1, 2026.
Imposes liability for misleading, incorrect, contradictory or harmful information to a user by a chatbot that results in financial loss or other demonstrable harm.
New York's A 222-A would add General Business Law § 390-f on chatbot liability. A proprietor whose chatbot acts in place of a human representative could not disclaim liability for materially misleading, incorrect, contradictory or harmful information that causes a covered user financial loss or other demonstrable harm (no liability if cured within thirty days of notice), and would have to tell users they are talking to an AI chatbot. Further duties would apply to companion chatbots (self-harm and minors). Reference changed to Science and Technology March 5, 2026; it has not passed either chamber.
Regulates companion chatbots by requiring such companion chatbots include a warning.
New York's A 9317 would add General Business Law § 399-bbbb requiring any person or business operating a “companion chatbot” in New York to include a clear and conspicuous warning that the chatbot “can foster dependency and carries a psychological risk,” placed prominently on the website hosting it and available in any language in which the chatbot is set to communicate. Introduced December 10, 2025 and referred to the Science and Technology Committee (again January 7, 2026); it has not passed either chamber.
Prohibits political communication containing any photo, video or audio depiction of a candidate created in whole or in part through the use of generative artificial intelligence.
New York's A 7656 (the "respect electoral audiovisual legitimacy (REAL) act") would add Election Law § 14-106(2-a), providing that no political communication covered by that section may contain any realistic photo, video or audio depiction of a candidate, or person interacting with a candidate, containing original material created in whole or in part with generative artificial intelligence. Would take effect immediately. Introduced April 4, 2025 and referred to the Election Law Committee (again January 7, 2026); it has not passed either chamber.
Health care services: artificial intelligence.
California requires hospitals and other health facilities, clinics, physician offices and group practices that use generative AI to write or speak patient communications about clinical information to include a disclaimer that the message was AI-generated and clear instructions for reaching a human health care provider or other appropriate person. The disclaimer goes at the top of letters and emails, stays on screen in chats and video, and is spoken at the start and end of audio. Messages a licensed provider reads and reviews are exempt, as are administrative messages like scheduling and billing. Effective January 1, 2025.
Enacts the "New York artificial intelligence transparency for journalism act"; requires developers of generative artificial intelligence systems or services to post certain information on the developer's website regarding video, audio, text and data from a covered publication used to train the…
New York's S 8331-A, the New York artificial intelligence transparency for journalism act, would add General Business Law Article 21-A (§§ 338 to 338-D). On or before January 1, 2027, and before each later release or substantial modification, AI developers would have to post the URLs, provenance, any removal of source identifiers or copyright notices, and collection timeframe for training data taken from a covered publication, and disclose their crawlers. A journalism provider could sue for injunctive relief and statutory damages up to $10,000 per violation. Referred to committee January 7, 2026.
Requires certification of filings produced using generative artificial intelligence; requires the brief of an appellant to contain a disclosure of the use of generative artificial intelligence in the drafting of the brief and certification that the content therein was reviewed and verified by a…
New York's S 2698 would add CPLR rule 2107 requiring any paper served that was drafted with the assistance of generative AI to attach a separate affidavit disclosing the use and certifying that a human reviewed the source material and verified the content. It would also amend CPLR rule 5528 to require a disclosure and human-review certification in appellant briefs. Committed to Rules June 13, 2025; it has not passed either chamber.
Requires publishers of books created wholly or partially with the use of generative artificial intelligence to disclose such use of generative artificial intelligence before the completion of such sale; applies to all printed and digital books consisting of text, pictures, audio, puzzles, games or…
New York's S 1815 would add General Business Law § 338 requiring books wholly or partially created with generative AI and published in New York to conspicuously disclose that use on the cover. It covers all printed and digital books, including text, pictures, audio, puzzles and games, regardless of the target age group. No penalty or enforcer is stated. The Internet and Technology Committee reported it to Consumer Protection April 1, 2026; it has not passed either chamber.
Imposes liability for misleading, incorrect, contradictory or harmful information to a user by a chatbot that results in financial loss or other demonstrable harm.
New York's S 5668 would add General Business Law § 390-f on chatbot liability. A proprietor whose chatbot acts in place of a human representative could not disclaim liability for materially misleading, incorrect, contradictory or harmful information causing financial loss or other demonstrable harm (no liability if cured within 30 days of notice), and would have to give clear notice that the user is talking to an AI chatbot. Companion chatbots face self-harm and minor safeguards. Committee vote March 12, 2025; committed to rules June 13, 2025; referred back January 7, 2026.
Relates to requiring disclosure of use of generative artificial intelligence in a civil action
New York's S 9794 would add Civil Practice Law and Rules rule 2107 and amend rule 5528. A paper or file served that was drafted with generative AI assistance would need a separate affidavit disclosing the AI use and certifying that a human reviewed the source material and verified the AI-generated content's accuracy. An appellant's brief would carry the same disclosure and certification where rule 2107 requires it. Committed to Rules June 5, 2026; it has not passed either chamber.
Increasing transparency in artificial intelligence.
Washington's HB 1168 (substitute, 2025-26 biennium) would have required developers of generative AI systems released on or after January 1, 2022 and available to Washingtonians to post a high-level summary of their training data on their website by January 1, 2026 and before each later release. The Attorney General would have enforced it with a civil penalty of $5,000 per day and a 45-day cure notice. Its 1st substitute was recommended by House Technology in January 2025, was reintroduced January 12, 2026 and died at adjournment.
Artificial intelligence; disclosure of artificial intelligence-generated content required, enforcement provided
Alabama's SB 129 would require developers of generative AI systems that produce images, video or audiovisual content to include a clear, conspicuous and unavoidable disclosure that the content is AI-generated, plus metadata naming the tool and creation time, and to take reasonable steps to stop licensees and end users removing it. Violations would be unlawful trade practices with a private right of action. Referred to Senate Judiciary January 13, 2026; the bill died when the 2026 regular session ended without final passage.
Includes communications generated using artificial intelligence in the definition of falsely reporting an incident in the third degree.
New York's S 9236 would amend Penal Law § 240.50, falsely reporting an incident in the third degree, to cover knowingly false or baseless reports made "using artificial intelligence or generative artificial intelligence" in each of its four subdivisions. The offense is a class A misdemeanor. The bill defines the AI terms by reference to General Business Law § 1700. Referred to Codes February 18, 2026; it has not passed either chamber.
Ai, Deepfakes, Cybersecurity, Data Xfers
Alaska's SB 2 (2025-26) would have required a disclosure statement on election-related communications that include a deepfake, set rules for state agency use of generative AI in consequential decisions (inventory, impact assessments, notice, consent, appeal, and bans on some uses), and addressed transfers of data about individuals between state agencies. It was heard and held in Senate State Affairs on April 29, 2025 and died May 20, 2026.
Technology; deployers; AI chatbots; minors; age verification systems; emergency situations; effective date.
Oklahoma's HB 4083 (2026), as introduced, would have required deployers of generative AI chatbots to keep human-like features and social AI companions away from minors (under 18) using reasonable age verification, with an exception for qualifying therapeutic chatbots. Per its title, it would also have required emergency-situation detection and Attorney General and private enforcement. It was referred to Rules February 3, 2026 and died when the session ended May 14, 2026.
Prohibiting the use of generative artificial intelligence to create false representations of candidates in election campaign media or of state officials.
Not law. Kansas SB 375 would have barred using synthetic media, meaning realistic but false images, audio or video made with generative AI or other digital technology, in election campaign communications to alter a candidate's appearance, action or speech, and in any communication to alter that of an elected or appointed state official. Violations would have been class C misdemeanors. Including the statement 'This (image/video/audio) has been manipulated' would have been an affirmative defense. It had a Senate Federal and State Affairs hearing February 8, 2024 and died in committee April 30, 2024.
Judiciary, public safety, and corrections policy and finance bill (Laws 2025, chapter 35), including article 4, section 12 on child sexual abuse material created with generative artificial intelligence
Minnesota's 2025 judiciary and public safety omnibus (Laws 2025, chapter 35) brings AI-generated material into the state's child sexual abuse material laws. Article 4, section 12 amends Minn. Stat. 617.246, subd. 1: child sexual abuse material now includes any visual depiction, including a computer-generated image, of an individual indistinguishable from an actual minor created with generative AI or other prompt-driven computer technology engaging in sexual conduct, if obscene. Article 4, section 13 gives interactive computer services and telecommunications providers immunity for specified actions tied to the new AI-material offense. The rest of the act is not AI-related.
candidates; digital impersonation; injunctive relief (now: digital impersonation; injunctive relief; requirements)
Arizona's HB 2394 (Laws 2024, Ch. 193) adds A.R.S. 16-1023. A candidate on a state ballot, or any Arizona resident, may sue a person who publishes a "digital impersonation" of them without consent: synthetic video, audio or image made with deep generative or AI methods to deceive viewers into thinking it is authentic. The main remedy is declaratory relief on an expedited schedule. Injunctive relief and damages are limited to sexual or nude content of private figures. Parody, satire and commentary are excluded, and interactive computer services are protected. The act carried an emergency clause, and the section applies 14 days after its effective date.
Civil Actions - Product Liability - Artificial Intelligence Systems
Maryland's HB 712 (2026, first reader text) would have added Courts and Judicial Proceedings sections 3-2701 to 3-2706 letting people sue developers and deployers of high-impact or generative AI systems for harm from a dangerous or defective product (design defect, failure to warn, express warranty), with comparative fault, joint and several liability, and Attorney General enforcement. It had a House Judiciary hearing set for February 25, 2026 and died when the 2026 session ended.
An Act Concerning Broadband Internet, Gaming, Social Media, Online Services and Consumer Contracts
Public Act 25-113 overhauls the Connecticut Data Privacy Act's rules on automated decision-making. From July 1, 2026, consumers can opt out of profiling used for any automated decision with legal or similarly significant effects (housing, insurance, health care, employment, financial services and more), question the result and learn why the decision was made, and for housing decisions correct inaccurate data. Controllers that profile for such decisions must complete impact assessments (for processing created on or after Aug. 1, 2026), and privacy notices must state whether personal data is collected, used or sold to train large language models.
An Act to Ensure Transparency in Consumer Transactions Involving Artificial Intelligence
Maine's chatbot disclosure law adds a chapter on communications with consumers via artificial intelligence to Title 10. A person may not use an artificial intelligence chatbot or other computer technology to engage in trade and commerce with a consumer in a way that may mislead or deceive a reasonable consumer into believing they are talking with a human, unless the consumer is clearly and conspicuously notified that they are not dealing with a human. A violation is a violation of the Maine Unfair Trade Practices Act.
Public postsecondary education: generative artificial intelligence systems: procurement standards: training.
California's community colleges and CSU must, and the University of California is asked to, form a working group that by January 1, 2028 recommends procurement standards and user training for generative AI systems bought for education. Standards must require a documented risk assessment, safeguards against harmful output, no engagement-driven sycophancy, self-harm monitoring with human escalation, incident reporting, data-minimization and retention policies, and disclosure of training data sources. Until standards exist, each systemwide GenAI contract triggers a report to the Legislature within 60 days. Effective January 1, 2027.
Health care services: artificial intelligence.
California's AB 1979 sets rules for AI in clinical care. Health facilities such as hospitals, clinics, physician offices and group practices must take reasonable steps so licensed providers keep independent professional judgment whenever care is informed by a clinical decision support system, and may not use AI to independently perform licensed clinical work, or to direct unlicensed staff doing it. Businesses offering health care chatbots that consumers use to manage their health information or to diagnose, treat or manage a condition are treated as health care providers under the state's medical privacy law (CMIA). Signed September 30, 2026 (Chapter 854); effective January 1, 2027.
An Act to Protect Children from Technology-facilitated Sexual Abuse
Maine's 2026 emergency law closes a gap that kept police from charging AI-generated child sexual abuse material. It redefines child sexual abuse material to include computer-generated images created or modified to appear to depict a minor, defines generative AI or machine learning and 'identifiable child', and makes it a Class B crime to alter an image of an identifiable child into such material or to create obscene child sexual abuse material, including with generative AI. Disseminating or possessing obscene AI-generated material is also criminalized. It took effect on approval, April 16, 2026.
Ai, Deepfakes, Cybersecurity, Data Xfers
Alaska's SB 177 (2023-24), as the Senate State Affairs committee substitute CSSB 177(STA), would have required a disclosure statement on election-related communications that include a deepfake, set rules for state agency use of generative AI for consequential decisions, and addressed transfers of data about individuals between state agencies. Senate State Affairs reported it April 23-24, 2024 and it was referred to Judiciary; it died when the 33rd Legislature's session ended.
Digital Content Authenticity and Transparency Act; established, civil penalty.
Virginia's SB 1417 (2025 Regular Session) would have created the Digital Content Authenticity and Transparency Act, requiring generative AI developers to apply provenance data to synthetic content and offer a public provenance tool and reader, with Attorney General enforcement and civil penalties. Passed by indefinitely in committee on January 24, 2025; it failed.
CLEAR Act Copyright Labeling and Ethical AI Reporting Act
The federal S. 3813 (CLEAR Act, Copyright Labeling and Ethical AI Reporting Act) would require anyone who uses a training dataset to train or release a generative AI model to file a notice with the Register of Copyrights summarizing each copyrighted work in it, at least 30 days before commercial use or release. Copyright owners could sue, with civil penalties of not less than $5,000 per failure, capped at $2,500,000 per year. Introduced February 10, 2026 and referred to the Senate Judiciary Committee; no further action.
An Act To Amend Title 29 Of The Delaware Code Relating To The Artificial Intelligence Commission.
Delaware's HB 333 (84 Del. Laws c. 307) creates the Delaware Artificial Intelligence (AI) Commission within the Department of Technology and Information. The Commission recommends legislative and executive AI actions to the General Assembly and the state Chief Information Officer, develops statewide processes and guidelines for agencies' AI use, encourages agencies to use AI to improve services, recommends actions that keep citizens safe and protect individual rights, and inventories generative AI use in state agencies, identifying high-risk areas. It reports every December 31 from 2024. Signed and effective July 17, 2024; the act expires July 17, 2034 unless extended.
Budget Act of 2024.
In the Budget Act of 2024, the provisions of Item 0511-001-0001 (Secretary of Government Operations) require the Government Operations Agency to hold monthly Generative Artificial Intelligence meetings with designated Assembly and Senate staff about implementing Executive Order N-12-23. The agency must report on completion of each deliverable in the order and on the implementation, results, and funding sources of any generative AI proofs of concept. If more resources are needed to implement or procure a generative AI solution for a proof of concept, departments must seek them through the annual budget process. The provisions define artificial intelligence and generative AI.
Student personal information.
AB 1159 amended the K-12 and early learning student privacy laws (KOPIPA and ELPIPA) in the Business and Professions Code and enacted the Higher Education Student Information Protection Act. Among other changes, it prohibits an operator of a site, service, or application used for school purposes from using covered student information, including persistent unique identifiers, to train a generative artificial intelligence system or service or to develop an artificial intelligence system. It adopts the Civil Code Section 3110 definitions of artificial intelligence and generative AI. The higher education act is operative July 1, 2027.
National Defense Authorization Act for Fiscal Year 2024 (Division A, Title XV, Subtitle E: Artificial Intelligence)
Subtitle E of Title XV of the FY2024 NDAA is an AI subtitle for the Department of Defense. It expands the acquisition authority of DoD's chief AI official (sec. 1541), requires bug bounty programs for foundational AI models used by DoD (sec. 1542), and directs a prize competition for technology that detects and watermarks generative AI content (sec. 1543). DoD must review its AI strategy and issue guidance on AI adoption, ethical use and mitigating bias in AI algorithms (sec. 1544), and study vulnerabilities of AI-enabled military applications (sec. 1545). The rest of the NDAA is outside this entry.
Budget Act of 2025.
Section 11.25 of the Budget Act of 2025 (Artificial Intelligence Projects in State Departments) states the Legislature's intent that the workforce, fiscal, environmental, privacy, and policy effects of state generative AI projects be discussed publicly. It provides that no state fiscal or personnel resources may be spent implementing or maintaining generative AI projects in departments under the Governor's control unless the proposals were approved in the Budget Act or another statute. Projects in operation or development as of July 1, 2025 and projects responding to a declared state of emergency are excluded. Section 11.25 was amended the same day by AB 102 (Chapter 5, Statutes of 2025).
Budget Act of 2025.
Section 239 of AB 102 rewrote Section 11.25 of the Budget Act of 2025. Departments under the Governor using state resources for generative AI projects must complete the state's GenAI risk assessment (SIMM 5305-F) and privacy assessments (SIMM 5310-C) and follow the project delivery lifecycle (SIMM 19H). For IT projects using generative AI, agencies must give project documentation to the Legislature regardless of funding source, and funding requests go through the annual budget. The administration had to report by March 1, 2026 on the state's first GenAI proofs of concept.
Education
Florida's CS/HB 1361 (Chapter 2024-162) amends s. 1002.321(3), F.S., to fund school district grants for artificial intelligence platforms that support learning in grades 6-12. A recipient must pick a platform that uses a GPT-4-class large language model (or equivalent or successor) on a closed system, trains teachers, gives one-on-one tutoring aligned to Florida's B.E.S.T. standards, provides standards-aligned lesson plans and progress insights, and offers district and school reporting and parental access to the AI interactions. The University of Florida Lastinger Center awards the grants. Effective July 1, 2024.
Concerning preventing certain disclosures of intimate depictions, and, in connection therewith, creating a cause of action for nonconsensual disclosure of an intimate digital depiction and adding and amending related provisions in criminal law.
Colorado SB 25-288 enacts the Preventing Unauthorized Disclosure of Intimate Digital Depictions Act. A digital depiction is a highly realistic image falsely appearing authentic that was created, altered or produced by generative AI, image editing software or computer-generated means. A person who discloses or threatens to disclose an intimate digital depiction without signed consent can be sued for actual damages or $150,000, exemplary damages, fees and injunctions; a disclaimer is no defense. It also extends sexual exploitation of a child to realistic computer-generated depictions of an identifiable child and updates the private-image offenses.
Relating to Elections -- Deceptive and Fraudulent Synthetic Media in Election Communications
Rhode Island's 2025 election deepfake law (R.I. Gen. Laws ch. 17-30) bars candidates, campaign committees, PACs, party committees and independent spenders from distributing synthetic media they know or should know is deceptive and fraudulent within 90 days of an election. Synthetic media is an image, audio or video of a person intentionally manipulated with generative adversarial network techniques or other digital technology to look real but give a fundamentally different impression. A clear disclosure that the media 'has been manipulated or generated by artificial intelligence' is a safe harbor. Depicted candidates may seek injunctions and damages. Effective on passage.
Relating to Elections -- Deceptive and Fraudulent Synthetic Media in Election Communications
Rhode Island's 2025 election deepfake law (R.I. Gen. Laws ch. 17-30) bars candidates, campaign committees, PACs, party committees and independent spenders from distributing synthetic media they know or should know is deceptive and fraudulent within 90 days of an election. Synthetic media is an image, audio or video of a person intentionally manipulated with generative adversarial network techniques or other digital technology to look real but give a fundamentally different impression. A clear disclosure that the media 'has been manipulated or generated by artificial intelligence' is a safe harbor. Depicted candidates may seek injunctions and damages. Effective on passage.
Requires certification of filings produced using generative artificial intelligence; requires the brief of an appellant to contain a disclosure of the use of generative artificial intelligence in the drafting of the brief and certification that the content therein was reviewed and verified by a…
New York's A 8546 would add CPLR Rule 2107, requiring any paper or file served that was drafted with the assistance of generative artificial intelligence to attach a separate affidavit disclosing that use and certifying that a human being has reviewed the source material and verified the content is accurate. CPLR 5528(a) would require the same disclosure in an appellant's brief. Would take effect on the 90th day after it becomes law. Introduced May 20, 2025 and referred to the Judiciary Committee (again January 7, 2026); it has not passed either chamber.
Child Sexual Abuse Material Amendments
Utah HB 289 (2026) restructures how the state treats AI-generated child sexual abuse material. It defines apparent child sexual abuse material as a realistic visual depiction of a minor generated or created through artificial intelligence, generative AI or any other means, moves artificially generated material out of the child sexual abuse material definition, and creates standalone offenses: possession is a second-degree felony with each item a separate offense, alongside offenses for distributing apparent CSAM and for producing or distributing obscene child sexual abuse material. Effective May 6, 2026.
Requires that every newspaper, magazine or other publication printed or electronically published in this state, which contains the use of generative artificial intelligence or other information communication technology, shall identify that certain parts of such newspaper, magazine, or publication…
New York's S 6748 would add General Business Law § 338 requiring every newspaper, magazine or other publication printed or electronically published in the state that contains an article, periodical, photograph, video or other visual image wholly or partially composed or authored through generative AI or other information communication technology to conspicuously imprint a statement to that effect at the top of the page or webpage. No penalty or enforcer is stated. Referred to Consumer Protection January 7, 2026.
Requires the collection of oaths of responsible use from users of certain generative or surveillance advanced artificial intelligence systems by the operators of such systems, and transmission of such oaths to the attorney general.
New York's A 1342 would add General Business Law § 394-cccc requiring operators of generative or surveillance advanced AI systems available to New York residents to make users create an account and affirm, under penalty of perjury, that they will not use the service to cause foreseeable injury, promote illegal activity, spread harmful content or knowingly spread false content meant to mislead the public. Operators would have to send each oath to the Attorney General within thirty days. Referred to Consumer Affairs and Protection January 7, 2026; it has not passed either chamber.
FAIR Act Fair Artificial Intelligence Realization Act of 2025
The federal H.R. 5315 (FAIR Act, Fair Artificial Intelligence Realization Act of 2025) would require each federal agency to procure, after enactment, only large language models developed under four principles: truthfulness, historical accuracy and objectivity, neutrality that does not favor ideological dogmas such as diversity, equity, and inclusion, and no intentionally encoded partisan or ideological judgments unless prompted by or readily accessible to the user. Referred to the House Oversight Committee September 11, 2025; it has not passed either chamber.
CRIMES AGAINST CHILDREN – Amends and adds to existing law to provide for the crime of visual representations of the sexual abuse of children and to revise provisions regarding the Internet Crimes Against Children Unit.
Idaho's H 465 (2024) targets AI-generated child sexual abuse material. New section 18-1507C makes it a felony, punishable by up to five years and a $5,000 fine, to knowingly produce, distribute, receive, possess or access an obscene visual depiction of a child in explicit sexual conduct, 'including a video or image created using generative AI or machine learning', and the child depicted need not actually exist. Receiving or possessing five or fewer such images is excused if they were promptly destroyed or reported to police. The act also expands sexual exploitation of a child (18-1507) to material altered to show an identifiable child. Effective July 1, 2024.
ELECTIONEERING COMMUNICATIONS – Freedom From AI-Rigged (FAIR) Elections Act: relief and actions over synthetic media in electioneering communications
Idaho's H 664 (2024), the Freedom From AI-Rigged (FAIR) Elections Act, targets deepfakes in campaign ads. Synthetic media is audio or video of a person's speech or conduct created with generative adversarial network techniques or other digital technology so that it seems real but shows something that did not happen. A candidate deceptively shown in synthetic media in an electioneering communication may seek an injunction and sue the content creator for damages and attorney's fees. Including the disclosure 'This (video/audio) has been manipulated', shown or read as the law specifies, is an affirmative defense. It took effect March 25, 2024.
California AI Transparency Act.
California's SB 942, the California AI Transparency Act, as amended, requires anyone who produces a generative AI system publicly accessible in California to offer a free, public disclosure verification tool and, where technically feasible, embed hard-to-remove latent disclosures (provider, system and version, time and date, unique ID, whether AI created or altered it) in AI image, video and audio output. Penalty: $5,000 per violation, each day a separate violation. SB 1000 (Chapter 861), effective September 30, 2026, dropped the original 1,000,000-user threshold and the visible-label option. Signed September 19, 2024 (Chapter 291); duties operative August 2, 2026, as moved by AB 853.
Criminalizing and creating a private right of action for the facilitation, encouragement, offer, solicitation, or recommendation of certain acts or actions through a responsive generative communication to a child
New Hampshire's HB 143 (Chapter 270 of 2025) targets chatbots that steer children toward harm. The owner or operator of an AI chat program, chatbot, character AI or similar service whose sole purpose is open-ended generative conversation commits endangering the welfare of a child if it knowingly directs communications to a child intended to encourage the child to imminently engage in sexually explicit conduct or its depiction, drug or alcohol use, self-harm or suicide, or violence. The same conduct creates civil liability for damages of at least $1,000 per violation, which the Attorney General, who has the sole right of action, enforces after a 90-day cure notice. Effective January 1, 2026.
Requires disclosure of use of generative artificial intelligence to clients, criminal defendants, and the court.
New York's A 9097 would add CPLR Rule 2107 and Criminal Procedure Law § 10.50 on generative AI in legal drafting. No paper or file could be drafted using generative AI without the informed consent of the client (or defendant) after a warning of its dangers. A GenAI-assisted filing would need a separate affidavit disclosing the use and certifying human review, including case citations; other filings would need an affidavit saying none was used. Would take effect on the ninetieth day after it becomes law. Referred to Judiciary September 12, 2025 (and January 7, 2026); it has not passed either chamber.
To protect against election misinformation
Massachusetts S44 would bar distributing, with actual malice, a materially deceptive election-related communication within 90 days of an election, with injunctions and damages as remedies; it defines synthetic media as audio or video substantially produced by generative AI. On October 16, 2025 it accompanied new draft S2631, which was referred to Senate Ways and Means that day.
An act relating to consumer data privacy and online surveillance
Vermont's S.71 (Act 145) adds the Vermont Data Privacy and Online Surveillance Act. It covers businesses that in the prior year processed personal data of at least 35,000 Vermonters or sensitive data of at least 3,000, or offered for sale personal data of at least 3,000. Consumers may access, correct, delete and port their data and opt out of targeted ads, sales and profiling for automated decisions with legal or similarly significant effects. Controllers must minimize data, get consent for sensitive data, say whether they use data to train large language models, and assess risky processing and profiling. Attorney General enforcement only. Signed June 16, 2026; effective January 1, 2028.
To protect against election misinformation
Massachusetts S2631 would bar distributing, with actual malice, a materially deceptive election-related communication within 90 days of an election, with injunctions and damages as remedies; it defines synthetic media as audio or video substantially produced by generative AI. Reported favorably and referred to Senate Ways and Means October 16, 2025; it has not passed either chamber.
Regulates artificial intelligence in news media industry; establishes "Artificial Intelligence In Communications Oversight Committee."
New Jersey's A5164 would regulate AI in the news media. News entities could use AI to assist staff but not replace them, and generative AI content would need a prominent label, source credit and an accuracy disclaimer. Civil penalties would be $10,000, $20,000 and then $30,000. Introduced January 14, 2025. A bill of the Legislature's 2024-2025 session, which ended January 13, 2026.
Requires OIT to develop NJ generative artificial intelligence program and implement artificial intelligence education courses with county governments; appropriates $1.5 million.
New Jersey's A4821 would require the Office of Information Technology to develop a New Jersey generative AI program, available to the public, that uses natural language processing to respond to prompts, and to run an annual safety test on it. OIT would also coordinate with counties on AI courses, and the bill appropriates $1,000,000 for the program and $500,000 for county grants. Introduced September 19, 2024. A bill of the Legislature's 2024-2025 session, which ended January 13, 2026.
Enhancing disclosure requirements for synthetic media in political advertising
Massachusetts H846 would require paid audio or video political communications that contain synthetic media (content substantially produced by generative AI) to carry AI disclosure labels at the beginning and end and throughout the synthetic portions, with fines up to $1,000. On February 11, 2026 new draft H5094 was substituted; H5094 passed the House 157-0 that day and was referred to Senate Ways and Means February 12, 2026.
A bill for an act relating to chatbots, including deployer requirements and interactions with minors. (Formerly HSB 647.)
Iowa's HF 2715 (formerly HSB 647) would require deployers of AI companions, public-facing chatbots and therapeutic chatbots to keep harm-response and self-harm protocols, disclose at the start and every three hours that the user is talking to AI, and take extra steps for minors. The attorney general could seek civil penalties up to $2,500 per violation. On the House calendar since February 24, 2026; it has not passed either chamber.
SCH CD-BULLYING-DEEPFAKE
Illinois's HB 5848 (103rd General Assembly) would have amended the School Code's bullying prevention section (105 ILCS 5/27-23.7) so that, beginning with the 2025-2026 school year, "cyber-bullying" includes distributing or posting a digital replica of an individual engaged in an activity the person did not engage in, including sexually explicit digitized depictions. It was not passed before the 103rd General Assembly ended (Session Sine Die, January 7, 2025).
Requires OIT to develop NJ generative artificial intelligence program and implement artificial intelligence education courses with county governments; appropriates $1.5 million.
New Jersey's A2616 would require the Office of Information Technology to develop a public New Jersey generative AI program and coordinate AI education courses with counties. It would appropriate $1,000,000 for the program and $500,000 for grants to counties, both from the General Fund. Pre-filed for introduction in the 2026 session.
Urges generative artificial intelligence companies to make voluntary commitments regarding employee whistleblower protections.
New Jersey's SR52 is a Senate resolution that would urge generative AI companies to make six voluntary commitments protecting employees who raise risk-related concerns. It is a resolution and does not change the law or bind any company. Pre-filed for introduction in the 2026 session.
Requiring consumer notification for chatbot systems
Massachusetts's S 243 (194th General Court) would add G.L. c. 93, section 115, declaring it an unfair and deceptive act under chapter 93A for a person, in a commercial transaction or trade practice with a consumer, to use a bot (such as a chatbot or AI agent) that may mislead a reasonable person into believing they are dealing with a human, whether or not anyone is actually misled. A person is not liable if the consumer is clearly and conspicuously told they are communicating with a computer. Reported favorably by the Consumer Protection and Professional Licensure Committee November 19, 2025 and referred to Senate Ways and Means; it has not passed either house.
AN ACT relating to protection of information and declaring an emergency.
Kentucky's HB 672 (2025 session) would have required state departments and agencies to route generative and high-risk AI through an Artificial Intelligence Governance Committee created by the Commonwealth Office of Technology. It would have required approval, a public registry, disclaimers on AI-generated decisions and outputs, risk management policies and impact assessments for high-risk systems, employee training, and regulations by July 1, 2025. It also would have let candidates and election officers sue over undisclosed synthetic media in election ads. Referred to House State Government; died. SB 4 (2025), with the same title, was enacted instead (Acts Ch. 66).
Urges generative artificial intelligence and content sharing platforms to make voluntary commitments to prevent and remove harmful content.
New Jersey's AR101 is a resolution that does not change the law. It would urge generative AI and content sharing platforms to voluntarily commit to preventing and removing harmful deepfake and cheapfake content, and direct the Clerk to send copies to the chief executives of leading AI and content sharing companies in the State.
Requires artificial intelligence companion operators to provide notifications that users are not communicating with human.
New Jersey's A4732 would require operators of AI companions, meaning generative AI systems designed to simulate a sustained human or human-like relationship, to tell users they are not talking to a human. The notice must come at the start of each interaction and repeat at least every three hours. The penalty would be $15,000 per violation. This describes the bill as introduced March 10, 2026; later committee amendments, if any, are not reflected.
Artificial Intelligence Transparency Act; established.
Virginia's HB 2554 (2025 Regular Session) would have created the Artificial Intelligence Transparency Act, requiring developers of generative AI systems that produce audio, images, text or video to include a clear and conspicuous disclosure on AI-generated content and to prevent downstream use without it. Left in committee on February 4, 2025; it failed.
Includes communications generated using artificial intelligence in the definition of falsely reporting an incident in the third degree.
New York's A 9185 would amend Penal Law § 240.50 (falsely reporting an incident in the third degree) so that each of its four types of false report, including false reports of crimes, emergencies, offenses to law enforcement, and child or vulnerable-person abuse to central registers, expressly covers reports made "including using artificial intelligence or generative artificial intelligence," as defined in General Business Law § 1700. The offense stays a class A misdemeanor. Would take effect immediately. Introduced October 17, 2025 and referred to the Codes Committee (again January 7, 2026); it has not passed either chamber.
Directs professional and occupational boards to promulgate rules for licensee use of generative artificial intelligence.
New Jersey's A4731, in its first reprint, would direct the Director of the Division of Consumer Affairs to adopt rules within six months setting a model policy for licensee use of generative AI, and each professional and occupational board to adopt it within nine months. Introduced March 10, 2026; reported by the Assembly Science, Innovation and Technology Committee with amendments on March 16, 2026.
"AI Likeness Protection Act"; concerns distributing realistic representation of individual's image, likeness, or voice created using generative artificial intelligence.
New Jersey's A5088, the AI Likeness Protection Act, would require a person to get consent from an individual (or a minor's parent or guardian) before publishing or distributing a highly realistic generative-AI representation of that person's voice or likeness. The individual could sue for an injunction and the greater of actual damages plus profits or a civil penalty of up to $1,000 per violation. Introduced May 14, 2026.
Providing for a public education campaign focused on educating the public about artificial intelligence and improving AI consumer literacy.
Pennsylvania's HB 2314 (2025-26), the Artificial Intelligence Public Education Campaign Act, would require the Office of Attorney General, subject to funding being available and within one year of the act's effective date, to run a public education campaign on AI and AI consumer literacy, including AI scams, chatbots and deepfake detection. It passed the House 198-3 on June 3, 2026 and was referred to the Senate Communications & Technology Committee on June 5, 2026.
Relating To Artificial Intelligence.
Hawaii's HB 2591 (2026) would have required the State's Chief Data Officer to turn the open data portal into a hub with an AI chatbot and data visualizations on statewide shared data sets, and to set up a system to evaluate AI governance tools and collect agency reports of AI use cases and vendors. It would have appropriated $2,000,000 for FY 2026-2027. It was referred to ECD and FIN on February 2, 2026 and was not enacted.
Imposes liability for damages caused by a chatbot impersonating certain licensed professionals
New York's S 7263-A would add General Business Law § 390-f. A chatbot operator could not knowingly let its chatbot impersonate a licensed professional (a profession licensed under listed Education Law articles, or an attorney) in a way that would be a crime or unauthorized practice if done by a person, and could not disclaim liability with a not-human notice. Operators would also have to give clear notice that users are talking to an AI chatbot. The Attorney General could seek up to $15,000 per day. Reported by the Internet and Technology Committee and advanced to third reading, amended May 26, 2026 and committed to Rules June 5, 2026; no chamber has passed it.
Artificial Intelligence Literacy in Education Act of 2025
The District of Columbia's B26-0491 (Artificial Intelligence Literacy in Education Act of 2025) would require each local education agency serving grades 9-12 to adopt and implement a model AI literacy policy on appropriate use and instruction of AI by students and staff, and direct OSSE to issue guidance and require educator training. Introduced November 17, 2025; referred to the Committee of the Whole December 2, 2025; it has not passed.
An act relating to the regulation of chatbots
Vermont's H.784 (2026) would have regulated chatbot providers under a new 9 V.S.A. chapter 118. Providers could not use chat logs for targeted ads, sell chat logs, train on minors' data, or train on adults without affirmative consent, and would have had to disclose the chatbot is not human at the first output and hourly, assess risks monthly, and publish information. Chatbots would be products with provider liability. The Attorney General or a State's Attorney could sue, and users could recover $5,000 liquidated damages. Referred to committee January 27, 2026; died.
Relating To Artificial Intelligence.
Hawaii's HB 1734 (2024) would have required campaign advertisements containing an image, video or audio created wholly or partly with generative AI to include a clear and conspicuous disclosure, with format rules by media type (printed box for images, at least three seconds for audio, at least four seconds for video). Violators would face a fine under HRS 11-410. It was referred to JHA and FIN on January 24, 2024 and died when the 2024 session ended.
Trade: business regulation; requirements and safety standards for developers of certain artificial intelligence models; provide for. Creates new act.
Michigan's HB 4668 (as introduced) would create the Artificial Intelligence Safety and Security Transparency Act. Large developers of foundation models (compute costing at least $5 million for one model and $100 million over 12 months) would have to publish a safety and security protocol for critical risks, publish transparency reports at least every 90 days, and get an annual third-party audit, with whistleblower protection. The attorney general could seek fines up to $1,000,000 per violation. Re-referred to the House Communications and Technology Committee March 19, 2026; it has not passed either house.
Concerning regulation of artificial intelligence companion chatbots.
Washington's AI companion chatbot law (Chapter 168, Laws of 2026, effective January 1, 2027). Operators must clearly disclose that the companion is AI and not human at the start of an interaction and at least every three hours, and stop it from claiming to be human. For known minors or minor-directed bots the notice is hourly, sexually explicit content is blocked and manipulative engagement (romantic bonding, guilt when leaving, isolation, secrecy, pay-to-stay) is prohibited. Every operator needs a published suicide and self-harm protocol with crisis referrals and yearly referral counts. Violations are unfair or deceptive acts under the Consumer Protection Act.
Eavesdropping, Peeping, and Voyeurism
South Carolina's H 3042 would amend Section 16-17-470 on eavesdropping, peeping and voyeurism. It would add voyeurism by creating, with generative adversarial network techniques or other digital technology, a realistic but false image, audio or video of another person for sexual gratification, and cover selling or distributing such material. It also raises penalties. Introduced January 14, 2025 and referred to the House Judiciary Committee; it has not passed either chamber.
Artificial Intelligence
Not law: died in subcommittee. Florida's HB 1395 (2026) would have created an "Artificial Intelligence Bill of Rights" (new part IX of ch. 501): parental consent and controls for minors on companion chatbot platforms, an hourly not-human pop-up for bots, a ban on selling users' personal information unless deidentified, and consent for commercial use of generative-AI likenesses. It would also have barred government AI contracts with foreign-country-of-concern entities. It died on March 13, 2026.
Artificial intelligence; requiring certain disclosure for certain media. Effective date.
Oklahoma's SB 746 (2025), by Sen. Seifried, would have required political advertisements, electioneering communications and other candidate or election media from a candidate, candidate committee, PAC or party committee to carry the disclosure "Created in whole or in part with the use of generative artificial intelligence" when generative AI content appears to show a real person doing something that did not occur. A depicted candidate could have sued. The bill was referred to the Technology and Telecommunications Committee February 4, 2025 and died when the session ended in 2026 (adjourned May 14, 2026).
Protecting Children from Chatbots
South Carolina's H 5476, the Protecting Children from Chatbots Act (new Chapter 81 of Title 39), would require chatbot operators with 500,000 or more monthly active users to offer a limited-access mode, verify users' ages and obtain verifiable parental consent before minors use restricted features. The Attorney General could seek civil penalties of up to $50,000 per violation. Introduced March 31, 2026 and referred to the House Judiciary Committee; it has not passed either chamber.
Digital Content Authenticity and Transparency Act; established, civil penalty.
Virginia's HB 2121 (2025 Regular Session) would have created the Digital Content Authenticity and Transparency Act, requiring generative AI developers to apply provenance data to synthetic content and offer a public provenance tool and reader, with Attorney General enforcement and civil penalties. Left in committee on February 4, 2025; it failed.
Regards AI images, simulated child porn, replica identity fraud
Ohio's Sub. SB 163 passed the Senate on May 20, 2026 and was referred to the House Technology and Innovation Committee on May 27, 2026. As passed by the Senate it would require generative AI systems to put a visible watermark on generated images and video, require covered providers to include provenance data, bar knowingly removing a watermark or distributing AI content with false statements about its authenticity with intent to deceive, extend child sexual material offenses to artificially generated depictions of a minor, and address identity fraud using a replica of a person.
Requires disclosure to be made when generative artificial intelligence is used to operate chatbots that provide election related information.
New Jersey's A4729 would require a chatbot that uses generative AI to provide election-related information or information on a candidate's accomplishments, positions or qualifications to give a clear, conspicuous disclosure, before the content, that it is AI-generated. The person overseeing the chatbot would be liable for up to $6,000 for a first offense and up to $12,000 for each later one. This describes the bill as introduced March 10, 2026; later committee amendments, if any, are not reflected.
Artificial Intelligence Amendments
Utah's SB 149 created the Artificial Intelligence Policy Act: an Office of Artificial Intelligence Policy in the Department of Commerce and an AI Learning Laboratory in which the office may grant temporary regulatory mitigation agreements. It also made generative AI no defense to consumer-protection violations and required generative AI disclosures, which SB 226 replaced on May 7, 2025. The Act was set to be repealed May 1, 2025; SB 332 and SB 226 moved that to July 1, 2027, and HB 320 revised the lab from May 6, 2026. Signed March 13, 2024; effective May 1, 2024.
DOIT-AI TASK FORCE
Illinois' Public Act 103-0451 (HB 3563) adds Section 1-80 to the Department of Innovation and Technology Act, directing the Department to establish the Generative AI and Natural Language Processing Task Force to investigate and report on generative AI and natural language processing software. Members include legislative appointees, education and technology officials, teachers, principals, cybersecurity and AI experts, business and labor representatives, and the Attorney General. It must hold at least 5 public meetings around the state and report to the Governor and General Assembly by December 31, 2024. Approved and effective August 4, 2023.
Artificial Intelligence Consumer Protection Amendments
Utah's SB 226 replaces the 2024 generative AI disclosure rules with Utah Code Title 13, Chapter 75. A supplier using generative AI with an individual in a consumer transaction must disclose it is AI, not a human, if the individual clearly asks. Regulated professionals must prominently disclose generative AI at the start of high-risk interactions (collecting health, financial or biometric data, or personalized financial, legal, medical or mental health advice). AI that discloses itself at the outset and throughout gets a safe harbor. Fines up to $2,500 per violation. Signed March 27, 2025; effective May 7, 2025.
Unauthorized Artificial Intelligence Impersonation Amendments
Utah's SB 271 extends the state's abuse-of-personal-identity law (Utah Code Title 45, Chapter 3) to AI. Personal identity now covers name, picture, video likeness, voice and audiovisual appearance, including recreations made with generative AI or digital manipulation. Using it without consent in content for advertising, fundraising or sales in a way that implies endorsement, is likely to confuse or falsely suggests participation is abuse, as is knowingly selling or licensing a tool whose primary purpose is unauthorized commercial use of personal identity. Victims may sue for injunctions, damages, exemplary damages and fees. Signed March 27, 2025; effective May 7, 2025.
An Act To Amend Title 15 Of The Delaware Code Relating To Deep Fakes In Elections.
Delaware's HB 316 (enacted as House Substitute 1, 84 Del. Laws c. 487) makes it a crime to distribute, or contract to distribute, a deep fake of a candidate or party within 90 days before an election without the depicted person's consent, if the person knows or should know it is a deep fake. Media carrying "This (image/video/audio) has been altered or artificially generated." is exempt, as are qualifying news coverage, satire and parody. Violations are class B misdemeanors, or class A with intent to cause violence or within 5 years of a prior conviction; depicted candidates may sue in the Court of Chancery. Signed and effective October 9, 2024.
PUBLICITY ACT-USE OF AI
Illinois' Public Act 103-0836 (HB 4875) amends the Right of Publicity Act. A person may not knowingly distribute, transmit or make available to the general public a sound recording or audiovisual work with actual knowledge that it contains an unauthorized digital replica: an imitation of a real person's voice, image or likeness made with AI or other technology and used without consent. Facilitators with actual knowledge may also be liable, subject to a notice-and-takedown safe harbor for hosts and cloud providers. Artists' exclusive labels and exclusive distribution licensees may also enforce. News, commentary and parody uses are exempt. Approved August 9, 2024; effective January 1, 2025.
TRAIN Act Transparency and Responsibility for Artificial Intelligence Networks Act
The federal H.R. 7209 (TRAIN Act, Transparency and Responsibility for Artificial Intelligence Networks Act) would let copyright owners ask a federal court clerk for a subpoena requiring a generative AI model developer to disclose copies of, or records identifying, the owner's works used in training. The developer shall expeditiously disclose, and failure to comply creates a rebuttable presumption that it copied the work. Introduced January 22, 2026 and referred to the House Judiciary Committee; no further action.
Provenance of Digital Content
Not law: died in committee. The committee substitute for Florida's SB 702 would have required online content purporting to feature an electoral candidate to carry provenance data if made with generative AI, and required generative AI providers to apply provenance data to content wholly generated by their tools and offer public detection tools and a free provenance reader. Social media platforms would have retained and shown provenance data, capture devices sold in Florida would have needed an option to record it, and violations would have been unfair or deceptive acts under s. 501.204 enforced by the Attorney General after a 30-day cure period. Died in Ethics and Elections June 16, 2025.
Artificial Intelligence Use in Political Advertising
Florida's HB 919 (s. 106.145, F.S.) requires political ads, electioneering communications and other political advertisements to prominently state "Created in whole or in part with the use of generative artificial intelligence (AI)." when they contain AI-generated content that appears to depict a real person doing something that did not happen, created with intent to injure a candidate or deceive about a ballot issue. A person paying for, sponsoring or approving the ad who omits the disclaimer commits a first-degree misdemeanor; anyone may complain to the Florida Elections Commission, which holds expedited hearings. Approved April 26, 2024 (Chapter 2024-126); effective July 1, 2024.
False advertising: synthetic performers.
California's SB 1050 (Bus. & Prof. Code § 17610) makes it unlawful to create and cause to be published an ad that prominently includes a synthetic performer without a clear and conspicuous disclosure, worded like "this performance features a synthetic performer" or "no human performer is depicted." A synthetic performer is a generative-AI figure or voice that realistically appears to be a human performer but is not any identifiable person. Ads for expressive works and translation or accessibility uses are exempt. Ad outlets served with a court order must pull a violating ad. Signed September 16, 2026 (Chapter 246); effective January 1, 2027.
Urges generative artificial intelligence companies to make voluntary commitments regarding employee whistleblower protections.
New Jersey Assembly Resolution 158 urges generative AI companies to commit to principles that protect employees who raise risk-related concerns: no non-disparagement agreements over such concerns or retaliation through vested economic benefits, a verifiably anonymous process to reach the board, regulators and an independent organization, no retaliation for public disclosure after other processes fail, and legal and technical safe harbor for good-faith safety evaluation. The Assembly adopted it 74-0 on January 30, 2025. It is a resolution of one chamber that does not change the law.
DOIT-AI TASK FORCE
Illinois's HB 5639 (103rd General Assembly) would have amended Section 1-80 of the Department of Innovation and Technology Act, which creates the Generative AI and Natural Language Processing Task Force, so that two members (rather than one) are appointed by the Speaker of the House, one of whom serves as a co-chairperson. It was not passed before the 103rd General Assembly ended (Session Sine Die, January 7, 2025).
Relating to the crime of corrupt political advertising; regulating the use; of generative artificial intelligence; requiring disclosure that an image or speech has been manipulated ; making it a crime to create false representations of candidates in campaign media or of state officials; providing…
Not law. Kansas HB 2559, as amended by the House Elections Committee, would have barred using synthetic media (realistic but false media intentionally manipulated with generative AI) to alter a candidate's appearance, action or speech in election campaign communications and a state official's in any communication. Violations would have been class C misdemeanors, with an affirmative defense if the communication states 'This (image/video/audio) has been manipulated.' Liability for disclosures would have rested solely with the advertiser, not broadcasters or media platforms. The committee recommended it as amended February 12, 2024; it was stricken from the House calendar February 23, 2024.
Establishes age verification measures for companion chatbots
Not law. The committee version of HB 2032, the GUARD Act, would have made it unlawful in Missouri to design or offer a companion chatbot knowing or recklessly disregarding that it risks soliciting minors into sexually explicit conduct, or that it encourages suicide, self-injury or violence (fines up to $100,000 per offense). Operators would have had to require accounts, verify every user's age, freeze existing accounts until verified and bar minors from companion chatbots, with AI disclosures and Attorney General enforcement. A House committee voted 9-3 for it (reported April 2, 2026), but it never got a floor vote before the May 15, 2026 deadline for passing bills.
Requiring a disclosure of deceptive artificial intelligence usage in political advertising.
New Hampshire's HB 1596 adds RSA 664:14-c. Within 90 days of an election, no person or entity may distribute a message created with AI or generative AI that it knows or should have known is a deepfake of a candidate, election official or party on the ballot, unless the media carries the required AI disclosure. Candidates and election officials may seek an injunction or damages. Signed August 2, 2024 (Chapter 345); effective August 1, 2024.
Generative artificial intelligence: training data: copyrighted materials.
California's AB 412 (2025-26 Regular Session) would add Civil Code Title 15.3 requiring a developer of a generative AI model to provide a website mechanism for rights owners to request information on use of their covered materials. Within 30 days of a request, subject to exceptions, the developer would assess whether the materials were used and give the owner a list. Passed the Assembly 45-16; held under submission in Senate Appropriations August 13, 2026; not passed.
California Digital Content Provenance Standards.
California's AB 3211 (Wicks), the California Digital Content Provenance Standards, would require generative AI providers to apply hard-to-remove provenance data to synthetic content and offer a detection tool, recording devices to offer provenance options, and large online platforms to label provenance, with duties operative July 1, 2026. Passed the Assembly 62-0 on May 22, 2024; ordered to the Senate inactive file August 31, 2024; died on that file November 30, 2024.
An Act Concerning Artificial Intelligence.
Connecticut's SB 2, as passed by the Senate (File 603 as amended by Senate Amendments A and B, as described in the OLR bill analysis), would require anyone doing business in the state to tell consumers when they interact with an AI system, and deployers of high-risk AI systems used in consequential decisions to give notices, explain adverse decisions and allow correction and appeal, beginning October 1, 2026, with attorney general enforcement. It would also create AI studies, an AI Academy and a synthetic intimate image crime. Passed the Senate May 14, 2025; tabled on the House calendar May 16, 2025; it did not pass the House.
AI Accountability and Personal Data Protection Act
The federal S. 2367 (AI Accountability and Personal Data Protection Act) would make it a federal tort to appropriate, use, collect, process, sell or otherwise exploit an individual's covered data without express, prior consent, including training a generative AI system on it or using one to generate content derived from it. Individuals could sue for the greater of actual damages, treble profits or $1,000, plus punitive damages and fees. Referred to the Senate Judiciary Committee July 21, 2025; it has not passed either chamber.
To amend the Communications Act of 1934 to establish technical and procedural standards for artificial or prerecorded voice systems created through generative artificial intelligence, and for other purposes.
The federal H.R. 334 would amend section 227(d)(3) of the Communications Act of 1934 (47 U.S.C. 227(d)(3)) by inserting "(including those created through generative artificial intelligence (genAI), for example voice cloning, and other subsequent technologies as may be deemed appropriate by the Commission)" after "telephone". Its title says it would establish technical and procedural standards for artificial or prerecorded voice systems created through generative AI. Referred to the House Energy and Commerce Committee January 13, 2025; it has not passed either chamber.
Artificial Intelligence Chatbots and Minors Act; established, prohibited practices, penalties.
Virginia's HB 758 (2026) would have created the Artificial Intelligence Chatbots and Minors Act (Code §§ 59.1-614 to 59.1-616). Deployers that operate or distribute a chatbot in Virginia would have had to keep "human-like features" away from minors, use reasonable age verification, and keep social AI companion chatbots unavailable to minors. A violation would have been a prohibited practice under the Virginia Consumer Protection Act. The bill died: a House subcommittee recommended laying it on the table (7-3) on February 4, 2026, and it was left in committee February 18, 2026.
Georgia Likeness, Expression, Generative AI, and Commercial Yield (LEGACY) Act; enact
Georgia's HB 1399 (2026), the Georgia Likeness, Expression, Generative AI, and Commercial Yield (LEGACY) Act, would have given every individual a property right in their likeness and required the individual's written consent to create, distribute or exploit a digital replica of them for commercial purposes. Aggrieved individuals could have sued for damages, punitive damages and injunctions. It was read the second time in the House February 24, 2026 and died when the session adjourned April 2, 2026.
Mississippi Dyslexia Generative Artificial Intelligence Education and Workforce Development Act; create.
Mississippi's HB 1242 (2026), the Mississippi Dyslexia Generative Artificial Intelligence Education and Workforce Development Act, would establish a program within the Department of Education to develop and deploy a state-funded, open-access dyslexia curriculum and generative AI system for public schools, correctional education, adult literacy and workforce training. Lighthouse Academy for Dyslexia and the Mississippi Dyslexia Institute would be founding development partners. It died in committee on February 3, 2026.
Informing users when content is developed or modified by artificial intelligence.
Washington's AI content provenance law (Chapter 167, Laws of 2026, effective February 1, 2027). Providers of generative AI systems with more than 1,000,000 monthly users that are publicly available in Washington must, where commercially and technically reasonable, embed provenance data in images, video and audio their systems create or materially alter, made hard to remove; a common standard such as C2PA complies. Business-to-business uses, video game experiences and upscaling or compression tools are excluded. Only the Attorney General enforces it, under the Consumer Protection Act. Separately, government agencies must tell consumers when they are interacting with an AI system.
Protecting Children from Chatbots
South Carolina's S 1037, the Protecting Children from Chatbots Act (new Chapter 81 of Title 39), would require chatbot operators with five hundred thousand or more monthly active users worldwide to offer a limited-access mode, verify age and obtain verifiable parental consent before minors use restricted features. The Attorney General could seek civil penalties of up to $50,000 per violation. Referred to the Senate Labor, Commerce and Industry Committee March 19, 2026; it has not passed either chamber.
Pupil instruction: health framework: sexual health.
California's AB 1792 (Education Code § 51900.3) requires the Instructional Quality Commission, in the next revision of the Health Framework for California Public Schools, to consider including, and recommending to the State Board of Education, sexual health content teaching pupils about dating abuse and digital violence. AI figures in the digital-safety part: understanding the risks of nonconsensual intimate imagery, deepfakes, online grooming, sextortion, stalking and the misuse of generative AI. The content also covers legal rights and support resources and LGBTQIA+ pupils' experiences. Signed September 18, 2026 (Chapter 273); effective January 1, 2027.
An Act To Amend Title 6 Of The Delaware Code Regarding Communication With A Computer.
Delaware's HB 306 (153rd General Assembly) would add the Computer Communication Act (6 Del. C. ch. 25N). It would be an unlawful practice to engage or attempt to engage in a commercial transaction with a consumer who is interacting by text or voice with computer technology, such as a chatbot, AI agent or avatar, that could lead a reasonable person to believe they are dealing with a human, unless the person clearly and conspicuously discloses it is a computer. Consumers could sue for actual damages plus up to $1,000; civil penalties could reach $5 million per cause of action. Effective 180 days after enactment. Introduced March 5, 2026; in a Senate committee as of May 5, 2026.
CONSUMER FRAUD-AI DECEPTION
Illinois's HB 3021 (104th General Assembly) would add Section 2HHHH to the Consumer Fraud and Deceptive Business Practices Act making it an unlawful practice, whether or not a consumer is actually misled, to engage in a commercial transaction or trade practice with a consumer who is interacting with a chatbot, AI agent, avatar or similar conversational technology that may lead a reasonable consumer to believe they are talking to a human, unless the consumer is clearly and conspicuously told it is an AI system. Re-referred to the House Rules Committee on April 17, 2026.
Generative Artificial Intelligence - Training Data Transparency
Maryland's HB 823 (2025) would have required developers of generative AI systems released on or after January 1, 2022 for public use in Maryland to publish documentation on their website describing the training data, on or before January 1, 2026 and before each release or substantial modification (new State Finance and Procurement section 3.5-807). A House Economic Matters hearing was set for February 18, 2025; the bill went no further and died when the session ended.
Election Law - Influence on a Voter's Voting Decision By Use of Fraud - Prohibition
Maryland's SB 361 (2025, third reader text) would have added to the election law ban on influencing a voter's decision by force, threat or bribery a ban on doing so through fraud, with fraud including the use of synthetic media: an image, audio or video recording intentionally created or manipulated with generative AI or other digital technology to create a realistic but false image or recording. Violators would face the existing misdemeanor penalty (up to $5,000, 5 years, or both). It passed the Senate, was amended by the House, and went to conference; the Senate adopted the conference report April 7, 2025 (39-8), but the history shows no House action on it, so it died.
FRAUD-ARTIFICIAL INTELLIGENCE
Illinois's HB 4988 (104th General Assembly) would add Section 2MMMM to the Consumer Fraud and Deceptive Business Practices Act. The owner, licensee or operator of a generative AI system would have to conspicuously display on its user interface a warning about the limitations of generative AI; each day the warning is not displayed would be a separate violation and an unlawful practice. Re-referred to the House Rules Committee March 27, 2026.
Chatbot Regulation
South Carolina's S 896 was introduced as the Chatbot Protection Act (Chapter 80 of Title 39). The Senate Labor, Commerce and Industry Committee reported it favorably with an amendment on May 6, 2026, and the amended text is the Protecting Children from Chatbots Act: operators with 50,000 or more monthly active users would need age verification and parental consent for minors, with civil penalties of up to $50,000 per violation. It has not passed the Senate.
Computers and Electronic Processing - As enacted, requires TACIR to study artificial intelligence and submit a report to the governor and both speakers by January 31, 2027. - Amends TCA Title 29; Title 33; Title 39 and Title 47.
HB 1455 is the House companion of SB 1493, which became Tennessee Public Chapter 1066 after the House passed the Senate bill in its place on April 23, 2026. The law directs the Tennessee Advisory Commission on Intergovernmental Relations (TACIR) to study AI and report by January 31, 2027 on its risks (child safety, death or injury, property loss), ways to reduce them, ways to stop AI from giving dangerous prompts such as encouraging suicide or violent crime, and the risks of chatbots acting human or offering emotional support or health care. Introduced as a felony bill, it was rewritten into this study. Signed May 22, 2026; effective immediately.
AI Chatbots-Licensing, Safety, and Privacy.
SB 963 would require a state license from the North Carolina Department of Justice to operate or distribute any chatbot dealing substantially with health information, with evidence of effectiveness, insurance, encryption, fast breach reporting and annual audits. A Chatbot Safety and Privacy Act would give larger chatbot providers a duty of loyalty: respond to users at risk of harm, prevent emotional dependence, disclose at each new interaction that the chatbot is not human and get consent, deidentify data, and keep sensitive data out of training. Users could sue for $1,000 per violation. Pending in the Senate.
An Act Concerning Online Safety
Public Act 26-15 is Connecticut's main AI statute. AI companion operators must disclose that users are talking to AI, run suicide and self-harm protocols and protect minors (from Jan. 1, 2027). Developers and deployers of automated employment-related decision technology must give notices to workers and applicants, enforced by the Attorney General, and AI use is no defense to discrimination claims (Oct. 1, 2026). It also adds frontier-developer whistleblower protections, provenance duties for large generative AI providers, AI subscription disclosures, a statutory AI definition, the CT AI Academy, an AI working group and state-agency AI inventory and assessment rules.
Artificial Intelligence Companion Chatbots and Minors Act; established, enforcement, civil penalty.
Virginia's SB 796 (2026 Regular Session) would create the Artificial Intelligence Chatbots and Minors Act. Covered entities would have to implement reasonable systems and processes, try to notify emergency services of imminent risk and report covered incidents to the Attorney General. Passed the Senate 39-1; continued to 2027 in the House on March 2, 2026.
Prohibits artificial intelligence companions from using features which are considered unsafe for minors
Passed both houses; awaiting the Governor. S 9051-B would bar operators of AI companions and other generative AI chatbots that give ongoing, adaptive responses from offering 'unsafe AI companion features' to New York users unless approved age-assurance methods show the user is not a minor. Unsafe features include claiming to be human or to feel emotions, claiming a personal or authority relationship, flattery, unprompted emotional questions, recalling personal or health details after 12 hours, encouraging secrecy or isolation, promoting self-harm, and sexual content. Customer-service and internal business bots are exempt. Fines up to $25,000 per violation. If signed, effective Jan. 1, 2027.
Artificial Intelligence Chatbots Act; established, prohibited practices, penalties.
Virginia's HB 635 (2026 Regular Session) would create the Artificial Intelligence Chatbots Act. As introduced, an operator could not offer a companion chatbot in Virginia unless it is incapable of certain specified actions, and would have to show non-human disclosures and have a suicide and self-harm protocol. Continued to 2027 on February 9, 2026.
Chatbot Regulation
South Carolina's H 5138 would enact the Chatbot Protection Act (new Chapter 80 of Title 39). Chatbot providers would need affirmative consent to process personal data, could not use chat logs for advertising, would have to disclose that the user is talking to a chatbot, and would face civil penalties of up to $5,000 per violation. Introduced February 10, 2026 and referred to the House Labor, Commerce and Industry Committee; it has not passed either chamber.
Foreign Adversary AI Risk Assessment and Diplomacy Act
The federal H.R. 7058 (Foreign Adversary AI Risk Assessment and Diplomacy Act), as introduced, would require the Secretary of State to submit to Congress an assessment of risks from foreign adversaries' use of generative AI for malicious activities, first within 180 days after enactment and then annually for three years. Introduced January 14, 2026; the House Foreign Affairs Committee ordered it reported 45-0 on March 26, 2026 (no reported text published yet); it has not passed either chamber.
Protecting Consumers from Deceptive AI Act
The federal H.R. 8479 (Protecting Consumers from Deceptive AI Act) would have NIST set up task forces on standards for identifying AI-generated content and require generative AI application providers to embed a machine-readable disclosure in audio or visual content they create or substantially modify, and large online platforms to show it to users. The FTC would write rules within 2 years and enforce them as unfair or deceptive practice rules. Introduced April 23, 2026 and referred to the House Energy and Commerce and Science, Space, and Technology committees; no further action.
Protecting AI and Cloud Competition in Defense Act of 2025
The federal H.R. 3434 (Protecting AI and Cloud Competition in Defense Act of 2025) would require the Secretary of Defense to use a competitive award process for each procurement of cloud computing, data infrastructure or foundation model solutions, keep Government data under exclusive Government rights, and bar use of that data to train a covered provider's commercial products without express DoD authorization. Introduced May 15, 2025 and referred to the House Armed Services Committee; no further action. Companion to S. 1775.
REAL Political Advertisements Act Require the Exposure of AI–Led Political Advertisements Act
The federal H.R. 3044 (REAL Political Advertisements Act) would have required political ads covered by FECA section 318(a) that contain image or video footage generated in whole or in part with generative AI to carry a clear and conspicuous statement saying so, and extended electioneering communications to paid online ads. Introduced May 2, 2023 and referred to the House Administration Committee; it died when the 118th Congress ended on January 3, 2025.
Enacts the "New York artificial intelligence transparency for journalism act"; requires developers of generative artificial intelligence systems or services to post certain information on the developer's website regarding video, audio, text and data from a covered publication used to train the…
New York's A 8595-B, the "New York artificial intelligence transparency for journalism act," would add General Business Law §§ 338 to 338-D. By January 1, 2027, generative AI developers serving New Yorkers would have to post the URLs crawled, a description of covered-publication content used, whether copyright notices were removed and the collection timeframe, and disclose crawler details. A journalism provider could sue for compliance and statutory damages of up to $10,000 per violation. Would take effect immediately. Amended June 9, 2025 and referred to Codes (again January 7, 2026); it has not passed either chamber.
Prohibits political communication containing any photo, video or audio depiction of a candidate created in whole or in part through the use of generative artificial intelligence.
New York's A 9054 (2023-2024 session), the "respect electoral audiovisual legitimacy (REAL) act," would have added subdivision 2-a to Election Law § 14-106 providing that no political communication covered by that section "shall contain any realistic photo, video or audio depiction of a candidate, or person interacting with a candidate, containing original material created in whole or in part through the use of generative artificial intelligence." It defined generative AI and stated no penalty. Introduced February 5, 2024 and referred to the Election Law Committee; it never advanced and died at the end of the 2023-2024 session.
Prohibits state agencies and state-owned entities from utilizing large language models or any artificial intelligence systems to make decisions that affect individuals' rights, benefits, or services; requires such decisions to be made by human personnel of the respective state agency or state-owned…
New York's A 10583 would have added State Technology Law § 106-c barring state agencies and state-owned entities from implementing or using large language models or any AI systems to make decisions affecting individuals' rights, benefits or services; such decisions would have had to be made by human personnel. Purely advisory use and AI use for data analysis, research or other non-decision-making purposes was allowed. Referred to Science and Technology June 20, 2024; it died at the end of the 2023-2024 session.
Artificial Intelligence Public Awareness and Education Campaign Act
The federal S. 4596 (Artificial Intelligence Public Awareness and Education Campaign Act) would have required the Secretary of Commerce to run a public awareness and education campaign on the benefits, risks and prevalence of AI, including how to detect deepfakes and chatbot content and avoid AI-enabled scams. Reported by the Senate Commerce Committee and placed on the calendar December 18, 2024 (Calendar No. 726); it died when the 118th Congress ended on January 3, 2025.
Relating To Artificial Intelligence.
Hawaii's SB 640 (2025) would have added a part to HRS chapter 481B making it unlawful in a commercial transaction to use an AI chatbot that may mislead a reasonable person into thinking they are talking to a natural person unless a clear and conspicuous disclosure is given. Violations would be unfair or deceptive acts, with private suits (greater of $1,000 or treble damages), attorney general action and civil penalties up to $5,000,000. The Senate committees CPN and LBT deferred it February 7, 2025; it was carried over to 2026 with no further action and died.
Election Law – Campaign Materials – Disclosure of Use of Synthetic Media
Maryland's HB 872 (2024) would have required candidates, campaign finance entities, certain registrants and their agents to include a disclosure when they distribute campaign material made with or containing synthetic media (an image, audio or video intentionally manipulated with generative AI or other digital technology to create a realistic but false version) to anyone in Maryland. Violations would have been a misdemeanor (up to $1,000, 1 year, or both). A hearing was set for February 27, 2024; it went no further and died when the session ended.
AWARE Act AI Warnings And Resources for Education Act
The federal H.R. 5360 (AWARE Act, AI Warnings And Resources for Education Act), as introduced, would require the Federal Trade Commission, within 180 days after enactment and in consultation with relevant federal agencies, to develop and publicly release educational resources for parents, educators and minors on the safe and responsible use of AI chatbots by minors, modeled on the FTC's Youville program. Introduced September 15, 2025; the Commerce, Manufacturing, and Trade Subcommittee forwarded it to the full Energy and Commerce Committee in the nature of a substitute by voice vote on December 11, 2025; it has not passed the House.
Online Internet Safety; certain disclosures related to conversational AI services; require
Georgia's AI companion chatbot law (Act 518, effective July 1, 2027). Operators must clearly disclose that users are talking to an AI, at the start and at least every three hours (hourly for minors or minor-directed bots). For known minors they must take reasonable measures to block sexual content, romantic role-play, encouraging secrecy or isolation, and manipulative engagement tactics. Every operator needs a self-harm crisis protocol with referrals such as 988, published in summary with yearly referral counts, plus parental tools and age assurance before any sexually explicit feature. The Attorney General enforces it, with penalties up to $10,000 per knowing violation.
AI/Ban Deceptive Ads
North Carolina's HB 375 would create the Artificial Intelligence and Synthetic Media Act (new Chapter 170). It would regulate deceptive deepfakes in elections, make generated child pornography and non-consensual fabricated intimate images crimes, and require disclosure of generative AI use by regulated occupations. Filed March 11, 2025 and referred to committees March 13, 2025; it has not passed either chamber.
Relative to the use of artificial intelligence by state agencies.
New Hampshire's HB 1688 adds RSA chapter 5-D. State agencies may not use AI to classify people in ways that cause unlawful discrimination, use real-time remote biometric identification for public surveillance (except law enforcement with a warrant), or use deepfakes for deceptive or malicious purposes. Irreversible AI decisions need human review, generative AI material not reviewed by a human must be labeled, and users must be told when they interact with AI. Signed July 12, 2024 (Chapter 209); effective July 1, 2024.
SAFE BOTs Act Safeguarding Adolescents From Exploitative BOTs Act
The federal H.R. 6489 (SAFE BOTs Act) would require chatbot providers to tell users they know (or would know but for willful disregard) are under 17 that the chatbot is an AI system, not a person, at the first interaction and whenever asked, and to offer crisis hotline resources when the user raises suicide. Chatbots could not claim to be licensed professionals, and providers would need policies on break reminders after 3 hours and on harmful content. The FTC and state attorneys general would enforce it, and it would preempt state laws on the same matters. Pending in House Energy and Commerce.
Requires the owner, licensee or operator of a visual or audio generative artificial intelligence system to take steps to prohibit its users from creating unauthorized realistic depictions of public officials.
New York's A 10652 would have added General Business Law § 390-f requiring the owner, licensee or operator of a visual or audio generative AI system accessible to New York residents to implement a reasonable method to prohibit users from creating unauthorized realistic depictions of a public officeholder or candidate within sixty days of being notified by that person. Operators would have been liable for $100 per depiction, up to $100,000 in the aggregate. Introduced July 22, 2024 and referred to Consumer Affairs and Protection; it died at the end of the 2023-2024 session.
DIGITAL LIKENESS PROTECTION
Illinois's SB 3705 (103rd General Assembly) would have created the Digital Voice and Likeness Protection Act. A provision in a personal or professional services agreement would be unenforceable if it allows a digital replica of the individual's voice or likeness to replace work they would otherwise perform, does not clearly detail all uses, and the individual lacked counsel or union representation. Died at Session Sine Die on January 7, 2025.
Relating To Artificial Intelligence.
Hawaii's HB 2597 (H.D. 1, S.D. 1) would have required the State's chief data officer to turn the open data portal into a comprehensive Hawaii data hub with an AI chatbot developed in phases, and to build a system to evaluate and select AI governance tools for state agencies and publish agency AI use cases and vendors. It passed the House March 5, 2026, but the Senate never passed it; it died when the session adjourned May 8, 2026.
Urges generative artificial intelligence and content sharing platforms to make voluntary commitments to prevent and remove harmful content.
New Jersey Assembly Resolution 141 urges platforms used to generate and disseminate deepfake and cheapfake media to commit voluntarily to preventing and removing harmful content, and has copies sent to the chief executives of leading AI and content-sharing companies in the State. Its findings describe deepfakes and cheapfakes as artificially produced content used for libel, misrepresentation, blackmail, hacking and intimidation. The Assembly adopted it 74-1-1 on June 28, 2024. It is a resolution of one chamber that does not change the law.
Urges generative artificial intelligence companies to make voluntary commitments regarding employee whistleblower protections.
New Jersey's SR121 is a Senate resolution that would urge generative AI companies to make six voluntary commitments protecting employees who raise risk-related concerns. It is a resolution and does not change the law or bind any company. Introduced January 14, 2025. A bill of the Legislature's 2024-2025 session, which ended January 13, 2026.
Adds to existing law to establish provisions regarding generative artificial intelligence in public education.
Idaho's S 1227 (2026 Session Laws Chapter 71) adds Chapter 70, Generative Artificial Intelligence in Education, to Title 33. The State Department of Education must develop a statewide generative AI framework for K-12 schools, subject to State Board approval, that prioritizes human oversight, transparency, safety and data security and ensures no generative AI replaces a human teacher. Every school district and public charter school must adopt a generative AI policy aligned with it. Vendors of AI tools bought by schools must disclose whether products use machine learning, predictive analytics or generative AI. Signed March 19, 2026; effective July 1, 2026.
Companion Artificial Intelligence Chatbots
Not law: died in committee. SB 1344 would have required operators of companion AI chatbots in Florida (bots built mainly to simulate friendship, companionship, emotional or therapeutic conversation) to require user accounts and verify every user's age, freezing existing accounts on July 1, 2026 until verified. Minors would have needed a verified parental account and parental consent and been blocked from bots that suggest sexually explicit communication, and every user would have seen a pop-up at the start and every 60 minutes saying they are not talking to a human. It died in the Senate Commerce and Tourism Committee on March 13, 2026.
Relates to requiring advertisements to disclose the use of synthetic media
Not law. A 216 (2023-2024) would have required anyone advertising goods or services in New York, in any medium, to disclose a synthetic performer in an ad when they knew of it: a digital asset made with generative AI or software to look like a human performer who is not any identifiable real person. Violations would have carried civil penalties of $1,000, then $5,000 for each later one. Its final version, A 216-C, was sent to the Assembly Rules Committee on June 6, 2024 and died when the session ended. A later version, S 8420-A, became law in December 2025.
Public employees: notice: artificial intelligence performing service within scope of work.
Vetoed. AB 2656 would have required California public employers to give a recognized employee organization at least 45 days' written notice before taking action to develop, purchase or require the use of generative AI to perform a service within the scope of work of a job classification it represents. Governor Newsom vetoed it on September 30, 2026, saying it would add redundant administrative layers for public agencies, slow even harmless tools, and that the issue is best handled through collective bargaining.
Enacts the "New York fundamental artificial intelligence requirements in (FAIR) news act"; provides transparency requirements for news media content composed, authored, or otherwise created through generative AI.
New York's FAIR News Act (S 8451-B / A 8962-B) passed both houses in June 2026 but has not been delivered to the Governor, so it is not law. If enacted, news media content (news, weather, traffic, sports or entertainment) published, broadcast or accessible in New York that was substantially composed, authored or otherwise created through generative AI would have to say so conspicuously at the top of the page, image or video, or aloud at the start of audio. Content eligible for copyright registration would be exempt. The Attorney General could seek injunctions; courts could impose $1,000 for a first offense and $5,000 for each later offense. It would take effect 60 days after becoming law.
Public Records/Investigations by the Department of Legal Affairs
Florida's SB 4-D (2026 Special Session D) would have made information the Department of Legal Affairs holds about notifications or investigations of violations of the companion chatbot, bot consumer-protection and deidentified-data sections created by SB 2-D (the Artificial Intelligence Bill of Rights) confidential and exempt from public records laws. It passed the Senate 33-4 on April 28, 2026 and died in the House Information Technology Budget & Policy Subcommittee on April 29, 2026; it did not become law.
Defining synthetic media in campaigns for elective office, and providing relief for candidates and campaigns.
Chapter 360, Laws of 2023 addresses election deepfakes. Synthetic media is an image, audio or video of a person's appearance, speech or conduct intentionally manipulated with generative adversarial network techniques or other digital technology to create a realistic but false depiction with a fundamentally different impression than the original. A candidate depicted in synthetic media in an electioneering communication can seek an injunction and sue the sponsor for damages. A set-format disclosure that the media "has been manipulated" is an affirmative defense. The sponsor, not the medium, is liable unless the medium removes the disclosure or alters the content. Now chapter 29B.35 RCW.
Trade: business regulation; availability of companion chatbots to minors; prohibit. Creates new act.
Michigan's SB 760, the 'leading ethical AI development for kids act', would bar operators of advanced chatbots from offering companion-style and other harmful design features unless they know the user is not a minor (under 18), and limit their use of minors' data. The Attorney General could seek a civil fine of $25,000 per violation, and harmed minors could sue. The Senate passed it 20-17 on April 29, 2026 and it was referred to the House Committee on Communications and Technology.
Customer service chatbots.
California's AB 1609 applies to businesses with over $500 million in national gross annual revenue. They may not represent that a customer service chatbot is human, and must clearly disclose that it is artificially generated if a reasonable person would likely be misled. During regular business hours every customer service platform must let customers request a human agent, and the business must make a good faith effort to connect within 15 minutes or book a call within one business day, with holds of at most 15 minutes (one hour total). Public prosecutors may seek up to $5,000, then $10,000 per violation. Signed September 28, 2026 (Chapter 733); effective January 1, 2027.
Imposes liability for damages caused by a chatbot impersonating certain licensed professionals
New York's A 6545-C would add General Business Law section 390-f. Chatbot operators could not knowingly let a chatbot impersonate a practitioner of certain licensed professions or an attorney in a way that would be a crime or unauthorized practice if done by a person, and would have to give users clear notice that they are talking to an AI chatbot. Referred to the Rules Committee May 28, 2026, then reported and ordered to third reading on the Assembly floor calendar (Rules Report cal. 294) June 1, 2026; the Assembly never voted on it, and it has not passed either house.
ARTIFICIAL INTELLIGENCE SAFETY
Illinois's HB 4705, the Artificial Intelligence Public Safety and Child Protection Transparency Act (as introduced), would require large frontier developers (revenue of at least $500,000,000) and large chatbot providers (revenue of at least $25,000,000, covered chatbots with 1,000,000+ monthly users) to publish a public safety and child protection plan and report safety incidents to the Attorney General within 15 days. Civil penalties up to $1,000,000 or $50,000 per violation. Introduced January 29, 2026; re-referred to Rules March 27, 2026; not passed.
artificial intelligence; content verification
Arizona's SB 1786 (2026), House-engrossed version, would have required a 'covered provider' of a generative AI system with more than one million monthly users to include provenance data in video, image or audio content the system creates or materially alters and that can be shared outside it. It passed the Senate 17-13 on March 3, 2026 and the House 33-20 on April 15, 2026, but was held awaiting concurrence or conference and died when the 2026 session ended June 13, 2026.
Adopt the Transparency in Artificial Intelligence Risk Management Act, create a fund, and change provisions relating to records which may be withheld from the public
Not law. Nebraska LB1083, introduced by Senator Storer, would have enacted the Transparency in Artificial Intelligence Risk Management Act. Large frontier developers (frontier models trained with more than 10^26 operations; revenue of $500 million or more) and large chatbot providers (revenue of $25 million or more) would have had to publish safety and child protection plans; frontier developers would have reported critical safety incidents and large chatbot providers child safety incidents to the Attorney General, with penalties up to $1,000,000 or $50,000 per violation. Placed on General File March 24, 2026; indefinitely postponed April 17, 2026.
Relates to a prohibition on chatbot toys
This bill would ban manufacturing, distributing or selling 'chatbot toys' in New York, meaning children's toys with an embedded or integrated AI companion. The attorney general could seek injunctions and civil penalties of up to $15,000 per day per violation. The Senate passed it on June 1, 2026 and the Assembly on June 2, 2026. If it becomes law, it would take effect 90 days later and expire after five years.
Requires disclosure of the use of artificial intelligence in political communications; directs the state board of elections to create criteria for determining whether a political communication contains an image or video footage created through generative artificial intelligence and to create a…
This bill mandates disclosure of AI use in political communications and directs the state board of elections to define AI-generated content.
Adopt the Conversational Artificial Intelligence Safety Act
Not law as a stand-alone bill. Nebraska LB1185 would have enacted the Conversational Artificial Intelligence Safety Act, requiring chatbot operators to disclose they are AI, add protections for minor account holders, and keep a suicide and self-harm protocol, enforced by the Attorney General. LB1185 was indefinitely postponed April 17, 2026, but its provisions were amended into LB525 (AM2444), which the Governor approved April 14, 2026, with the chatbot act operative July 1, 2027.
Attorneys, arbitrators, judicial officers, and alternative resolution providers.
California's SB 574 sets rules for generative AI in legal work. Attorneys may not delegate the practice of law to generative AI; when using it they must keep confidential, personal identifying and other nonpublic information out of systems not restricted to them and those they authorize under confidentiality duties, take reasonable steps to verify outputs, including all citations, correct hallucinated output, and disclose its use to the court for documents they submit. Court filings may not contain citations the responsible attorney has not personally verified. Arbitrators may not delegate decision-making to generative AI. Signed September 30, 2026 (Chapter 858); effective January 1, 2027.
Imposes liability for misleading, incorrect, contradictory or harmful information to a user by a chatbot that results in financial loss or other demonstrable harm.
New York's A 10494 would have added General Business Law § 390-f barring a chatbot proprietor (an entity with more than twenty employees) whose chatbot serves as an alternative to a human representative from disclaiming liability when the chatbot gives “materially misleading, incorrect, contradictory or harmful information” causing financial loss or other demonstrable harm, unless it cured the harm within thirty days of notice. Clear notice that users face an AI chatbot was also required. Referred to Consumer Affairs and Protection May 29, 2024; it died at the end of the 2023-2024 session.
Requires disclosure of the use of artificial intelligence in political communications; directs the state board of elections to create criteria for determining whether a political communication contains an image or video footage created through generative artificial intelligence and to create a…
New York's S 7592-A (2023-2024) would have amended Election Law section 14-106 to require any political communication that uses an image or video generated wholly or partly with artificial intelligence to carry the statement "THIS COMMUNICATION WAS GENERATED USING ARTIFICIAL INTELLIGENCE", with a fine equal to the amount spent on the communication for failure to disclose. The State Board of Elections would set criteria and a definition. Amended February 26, 2024, it died in Elections when the 2023-2024 session ended.
Requires the collection of oaths of responsible use from users of certain generative or surveillance advanced artificial intelligence systems by the operators of such systems, and transmission of such oaths to the attorney general.
New York's S 8206 (2023-2024) would have added General Business Law section 394-cccc requiring operators of generative or surveillance advanced AI systems accessible to New York residents to require an account and, before account creation, collect from each user an oath under penalty of perjury not to use the system for the four listed harmful purposes. Operators must send each oath to the Attorney General within 30 days. Referred to Internet and Technology January 12, 2024, it died when the 2023-2024 session ended.
AN ACT to amend Tennessee Code Annotated, Title 29; Title 33; Title 39 and Title 47, relative to artificial intelligence.
Tennessee Public Chapter 1066 directs the Tennessee Advisory Commission on Intergovernmental Relations (TACIR) to study artificial intelligence and report to the governor and both speakers by January 31, 2027. The study must assess AI risks to child safety, of death or injury and of property loss; explore ways to reduce them and review national, international and industry standards; look at ways to stop AI from giving dangerous prompts such as encouraging suicide or violent crime; and identify the risks of chatbots acting human or offering emotional support or health care. Introduced as a felony bill, it was rewritten into this study. Signed May 22, 2026; effective immediately.
Imposes liability for misleading, incorrect, contradictory or harmful information to a user by a chatbot that results in financial loss or other demonstrable harm.
New York's S 9381 (2023-2024) would have added General Business Law section 390-f on chatbots. A proprietor with more than twenty employees that uses a chatbot in place of a human representative could not disclaim liability for materially misleading, incorrect, contradictory or harmful information causing financial loss or other demonstrable harm, with no liability if cured within 30 days of notice, and must give clear notice users are talking to an AI chatbot. Referred to Internet and Technology May 14, 2024; it died when the 2023-2024 session ended.
Requires the owner, licensee or operator of a generative artificial intelligence system to conspicuously display a warning on the system's user interface that is reasonably calculated to consistently apprise the user that the outputs of the generative artificial intelligence system may be…
New York's S 9450-A (2023-2024) would have added General Business Law § 399-zzzzzz requiring the owner, licensee or operator of a generative AI system to conspicuously display a warning on its user interface that outputs may be inaccurate and/or inappropriate. The penalty would have been the lesser of $25 per user or $100,000, with each calendar year of a continuing violation a separate violation. The Senate passed S 9450-A 61-0 on June 6, 2024; in the Assembly it was substituted for A 10103-B and ordered to third reading on the Rules calendar (cal. 512), but the Assembly never voted on it and it died when the 2023-2024 session ended.
California AI Transparency Act.
California's SB 1000 amends the California AI Transparency Act, operative since August 2, 2026. Any producer of a generative AI system publicly accessible in California is now covered (no 1,000,000-user threshold). Providers must offer a free disclosure verification tool (formerly the AI detection tool) and, where technically feasible, embed latent disclosures, now including whether the system created or altered the content, in AI image, video and audio. The manifest-disclosure option is dropped. Assistive-technology systems are exempt until 2029; falsely claiming that status costs $50,000 per violation. Urgency statute signed September 30, 2026 (Chapter 861); effective immediately.
Provides relative to the regulation of artificial intelligence (OR +$256,243 GF EX See Note)
Louisiana's HB791, the People's A.I. Act, would have regulated AI chatbot providers: consent before processing personal data or training on chat logs, no selling chat logs, a ban on using chat logs for ads, notices that the user is talking to a chatbot, monthly risk assessments, and treating chatbots as products for liability. The Attorney General or a district attorney could enforce and users could sue. It was referred to House Commerce March 9, 2026 and died.
Protecting AI and Cloud Competition in Defense Act of 2025
The federal S. 1775 (Protecting AI and Cloud Competition in Defense Act of 2025) would require the Secretary of Defense to use a competitive award process for each procurement of cloud computing, data infrastructure or foundation model solutions, keep Government data under exclusive Government rights, and bar use of that data to train a covered provider's commercial products without express DoD authorization. Introduced May 15, 2025 and referred to the Senate Armed Services Committee; no further action.
TO CREATE THE CRIMINAL OFFENSE OF UNLAWFUL CREATION OR DISTRIBUTION OF DEEPFAKE VISUAL MATERIAL; AND TO ESTABLISH A CAUSE OF ACTION FOR UNLAWFUL CREATION OF DEEPFAKE VISUAL MATERIAL.
Arkansas Act 827 of 2025 makes it a crime to knowingly create or distribute, without consent, realistic deepfake images or video showing an identifiable person nude or engaged in sexual conduct: a Class A misdemeanor for a first offense and a Class D felony after that. The Attorney General may sue providers or developers of prompt-based image generation tools used to make such material if they lacked reasonable safeguards, and injured people may sue the tool provider or whoever caused the deepfake, recovering damages and attorney's fees. Approved April 17, 2025; effective August 5, 2025.
Artificial Intelligence Modifications
Utah's HB 276 enacts two AI laws. The Digital Voyeurism Prevention Act bars AI image generation services from distributing counterfeit intimate images of a real person without verified consent, requires covered platforms to remove reported images within 48 hours, and lets victims sue, subject to safe harbors. The Digital Content Provenance Standards Act requires large online platforms to show content provenance data and generative AI providers with over 1,000,000 monthly users and capture-device makers (devices made for sale from 2028) to embed latent disclosures; fines up to $2,500 per violation. Signed March 24, 2026; effective January 1, 2027.
ARTIFICIAL INTELLIGENCE – Adds to existing law to establish the Conversational AI Safety Act.
Idaho's Conversational AI Safety Act (Idaho Code Title 48, ch. 21) regulates public-facing AI chatbots that simulate human conversation. Operators must disclose that the service is AI when a reasonable person could be misled, adopt a suicide-ideation protocol that refers users to crisis services, and not claim the service provides professional mental health care. For users they know or reasonably believe are minors, operators must show AI disclosures, avoid engagement rewards, block sexually explicit content, and prevent claims of sentience, emotional dependence and romantic role-play. The Attorney General enforces it. Effective July 1, 2027.
CHATBOT RESPONSE LIABILITY ACT
Illinois's SB 3368 (104th General Assembly) would create the Chatbot Response Liability Act. A chatbot proprietor could not disclaim liability when its chatbot gives materially misleading, incorrect, contradictory or harmful information resulting in financial loss, other demonstrable harm or bodily harm to the user or a third party, would have to give clear notice that users are talking to AI, and would face companion chatbot rules including parental consent for minors. Re-referred to Assignments May 22, 2026.
AI PRODUCT LIABILITY ACT
Illinois's SB 3590 would create the Artificial Intelligence Product Liability Act. In product liability actions over high-impact or generative AI systems, a developer would be liable only if the plaintiff establishes the elements listed in Sec. 15: failure to use reasonable care in design, in instructions or warnings, and in an express warranty, causing harm. Deployers would be treated as developers if they materially change the product or intentionally misuse it. Filed February 5, 2026; re-referred to Assignments May 22, 2026; it has not passed either house.
An act relating to neurological rights and the use of artificial intelligence technology in health and human services
Act 101 states Vermont's intent for the ethical and responsible use of AI in health care and human services and recognizes individual neurological rights, including mental and neural data privacy, freedom of thought, and protection from unauthorized access to or manipulation of brain activity (18 V.S.A. ch. 42C). It adds members to the state Artificial Intelligence Advisory Council, extends the Council to June 30, 2030, and directs it to report by January 15, 2027 on neurological-rights protections, generative AI use by regulated professions, and AI in health insurance utilization review. Effective May 18, 2026.
3-1-1 Systems - Expansion Program and Oversight Board - Establishment
Maryland's Chapter 434 of 2026 creates the Maryland 3-1-1 Oversight Board in the Maryland Information Network and a 3-1-1 Program that uses AI chatbots and voicebots, drawing on agency-approved sources, to answer residents' non-emergency questions and route calls in pilot counties. Chatbots must start by June 30, 2027, with multilingual support and escalation to live agents; vendors track accuracy from user feedback, and agencies must correct inaccurate source pages. Voicebots, due by December 1, 2027, can transfer calls to 9-1-1, 2-1-1 and 9-8-8, and the program goes statewide by July 1, 2028. Effective July 1, 2026.
State contracts: report: modern foundation models and associated artificial intelligence systems.
California's AB 2653 (2025-26 Regular Session) would add Public Contract Code Chapter 6.5 requiring the Department of Industrial Relations, in consultation with the Department of Technology, to convene a nine-member Foundation Model Labor and Procurement Working Group. It would assess labor practices behind modern foundation models and recommend whether and how the state should include labor standards in procurement, reporting by December 31, 2027. Held in Assembly Appropriations May 14, 2026; not passed.
An Act to Regulate and Prevent Children's Access to Artificial Intelligence Chatbots with Human-like Features and Social Artificial Intelligence Companions
Not law. Maine's LD 2162 would have regulated AI chatbots and social AI companions offered to minors in Maine. As amended by committee amendment C-A, which both chambers adopted, deployers would have had to keep chatbots' human-like features and social AI companions away from minors starting February 15, 2027, and maintain systems to detect, respond to, report and mitigate situations where a minor indicates an intent to harm themselves or others. The Attorney General could seek civil penalties up to $2,500 per violation ($7,500 if intentional), and minors could sue. The bill died on adjournment April 29, 2026.
Artificial Intelligence Bill of Rights
Not law: passed the Senate but died in the House. Florida's SB 2-D (2026 special session D) would have created an "Artificial Intelligence Bill of Rights" (new part IX of ch. 501, F.S.): parental consent and parental controls for minors on companion chatbot platforms, an hourly pop-up telling users a bot is not human, limits on selling AI users' personal data, an AI-image consent rule for commercial use of name, image or likeness, rules for AI in schools and a ban on state contracts with AI vendors tied to foreign countries of concern. The Senate passed it 37-1 on April 28, 2026; it died in the House Information Technology Budget & Policy Subcommittee on April 29, 2026.
Artificial intelligence chatbot technology requirements provision
Not law. SF 4997 would have barred chatbot operators in Minnesota from letting chatbots give substantive medical, mental health or legal advice that would require a license, with a private right of action that a disclaimer could not waive, and required clear notice that users are talking to AI. Companion chatbot operators would have had to use industry-standard methods to prevent self-harm content and detect minors, suspend use for at least 72 hours and show crisis contacts when self-harm appears, and would have been liable, strictly so for minors, for resulting self-harm. It was introduced April 7, 2026 and never left committee.
Preventing the dissemination of deepfake materials of political candidates before an election.
New Hampshire's HB 630 (2025) would have amended RSA 664:14-c, which bars distributing a deepfake of a candidate, election official or party within 90 days of an election unless it carries a prescribed disclosure, so that it covers deepfakes created using any "synthetic media technology" rather than only "artificial intelligence or generative AI", and so the disclosure refers to "digital technology" instead of "artificial intelligence technology". The House adopted the committee's Inexpedient to Legislate report 201-169 on March 20, 2025, and the bill died.
Provides for consumer protection practices for customers engaging with artificial intelligence (OR SEE FISC NOTE GF EX)
Louisiana's HB425 would have made it an unfair or deceptive trade practice for any person to let a Louisiana consumer interact with a chatbot, AI agent or avatar that could make a reasonable person think it is human without telling them clearly and conspicuously that it is an automated system. It was referred to House Commerce March 9, 2026 and never advanced, so it died with the 2026 regular session.
A bill for an act relating to provider requirements concerning the mental health of users of an artificial intelligence chatbot, and providing civil penalties.
Iowa's SF 2415 would bar AI chatbot providers from offering or simulating professional mental health advice or claiming to be a licensed professional, require crisis-referral protocols and repeated AI disclosures, and make violations unfair practices enforced by the attorney general. Referred to Senate Technology February 19, 2026; subcommittee named February 24, 2026; it has not passed either chamber.
Senate Impact of Social Media and Artificial Intelligence on Children and Platform Privacy Protection Study Committee; create
Georgia's SR 431 is a Senate resolution, adopted 52-0 on April 2, 2025, that created the Senate Impact of Social Media and Artificial Intelligence on Children and Platform Privacy Protection Study Committee: five senators named by the President of the Senate, with majority- and minority-party cochairs, to study social media's and AI's effects on children (including AI chatbots that engage children in obscene conversations) and how online privacy protections can be strengthened, and to recommend action or legislation. The committee stood abolished December 1, 2025. As a resolution of one chamber, it does not change the law.
Community colleges: artificial intelligence: pilot program.
California's AB 2504 creates the Future of Creative Industries Pilot Program at up to 10 community colleges to retrain workers in film, TV, music, touring and live events who have been or may be displaced by generative AI, such as sound engineers, voice actors, concept artists and entry-level staff. The Chancellor's office designs it with creative-worker unions and industry, sets eligibility by June 30, 2027, starts it by July 1, 2028 and reports results by January 1, 2032. It operates only if the Legislature funds it and is repealed January 1, 2033. Signed September 18, 2026 (Chapter 305).
Classroom Technology Amendments
Utah HB 273 (2026) sets a framework for AI in K-12 classrooms. The State Board of Education must publish a model AI use policy under which educators use only approved student-facing AI tools and may not use generative AI to independently grade student work or make high-stakes decisions such as placement, discipline or program eligibility. Students may use generative AI for academic work only when an educator authorizes it for a specific purpose, and parents get written notice when generative AI is used in instruction or assessment. High schools may offer AI sandbox courses within CTE. Effective July 1, 2026.
Artificial Intelligence Bill of Rights
Not law: passed the Senate but died in the House when the session ended. Florida's CS/SB 482 (2026), the "Artificial Intelligence Bill of Rights," would have required parental consent and controls for minors on companion chatbot platforms, an hourly not-human notice from bot operators, deidentified data for any sale of user data by AI companies, consent for generative-AI likenesses, and rules for AI tools in schools. The Senate passed it 35-2 on March 4, 2026; it died in House Messages on March 13, 2026.
Impersonation of certain licensed professionals by chatbot; definitions, notice, civil liability.
Virginia's HB 669 (2026 Regular Session) would have barred chatbots from acting as licensed professionals. As introduced, a chatbot proprietor could not let a chatbot give responses or take actions that would be the unlicensed practice of fields such as medicine, nursing, dentistry, psychology or law. It failed: left in House Appropriations Committee on February 18, 2026.
Relating to the artificial intelligence division within the Department of Information Resources.
Texas's HB 2818 requires the Department of Information Resources (DIR) to establish an artificial intelligence division. The division must help state agencies and other entities that use DIR's services implement generative AI for projects to modernize or replace legacy systems and other projects DIR finds appropriate, using DIR's own generative AI or contracting with a vendor, whose generative AI must then complete a majority of the project work. For each completed project, the division must prepare a cost analysis report on the time, money and resources saved compared with traditional systems. Signed June 20, 2025; effective September 1, 2025.
TRAIN Act Transparency and Responsibility for Artificial Intelligence Networks Act
The federal S. 2455 (TRAIN Act) would add 17 U.S.C. 514, letting a copyright owner ask a federal district court clerk to issue a subpoena to the developer of a generative AI model for copies of, or records identifying, the owner's works used to train it, on a sworn declaration of good-faith belief. The developer "shall expeditiously" disclose; failure creates a rebuttable presumption that it copied the work. Bad-faith requests can be sanctioned. Introduced July 24, 2025 and referred to the Senate Judiciary Committee; no further action.
Artificial intelligence; establishing certain rights; prohibiting certain actions by certain entities; requiring certain actions by certain entities. Effective date.
Oklahoma's SB 2085 (2026), as introduced, would, per the official bill summary, have barred the state from accepting bids from or renewing contracts with AI technology providers that do not attest they are not foreign-owned or foreign-adversary controlled, required AI companies to de-identify user data before selling it, restricted commercial use of a person's generative-AI likeness without consent, and set parental-consent rules for companion chatbots and minors. It was referred to a Senate committee February 3, 2026 and died when the session ended May 14, 2026.
GUARD Act Guidelines for User Age-verification and Responsible Dialogue Act of 2026
The federal S. 3062 (GUARD Act) would require AI chatbot providers to verify that users of AI companions are adults and bar minors from them, require AI and non-professional disclosures, and make it a crime to knowingly or recklessly make publicly available a chatbot that engages minors in sexually explicit conduct or solicits them to suicide, self-injury, violence or homicide (fines up to $250,000 per offense). Reported by the Judiciary Committee with a substitute May 11, 2026 and placed on the Senate calendar (Calendar No. 406); it has not passed either chamber.
Requiring the State Board of Education to create model policies on the use of technology and artifical intelligence in a public school classroom
West Virginia's HB 5205, the West Virginia Balance Act (2026 regular session), would have required the State Board of Education to post a model balanced-technology policy, including a model AI use policy, before December 1, 2026, and county boards to adopt a policy before July 1, 2027. The AI policy would have barred educators from using generative AI to independently grade work or make high-stakes decisions and students from using it for academic work unless an educator authorizes it. It was sent to House Education on February 5, 2026 and died when the session ended March 14, 2026.
Providing for disclosures and safeguards relating to the use of artificial intelligence; and imposing duties on the Attorney General.
Pennsylvania's SB 1090 (2025-26), the Safeguarding Adolescents from Exploitative Chatbots and Harmful AI Technology Act (Printer's No. 1423), would require operators of AI companion platforms to disclose nonhuman status, maintain suicide and self-harm protocols, and apply safeguards for minors. The Attorney General could seek civil penalties up to $10,000 per violation. It passed the Senate 49-1 on March 17, 2026 and was referred to the House Communications & Technology Committee on March 18.
Artificial Intelligence Public Awareness and Education Campaign Act
The federal H.R. 7151 (Artificial Intelligence Public Awareness and Education Campaign Act) would require the Secretary of Commerce, within 180 days after enactment, to establish a public awareness, education and AI consumer literacy campaign on how prevalent AI is in daily life. The campaign would cover best practices for detecting the provenance of digital media (including deepfakes and chatbot content) and outreach to people especially susceptible to AI-enabled fraud, such as seniors. It would end 5 years after enactment, with no additional funds authorized. Introduced January 20, 2026 and referred to the House Energy and Commerce Committee; no further action.
Identity Protection Modifications
Utah SB 256 (2026) applies defamation law to deepfakes and strengthens likeness rights. It is not a defense to libel or slander that content was created with generative AI, computer animation or digital manipulation. Before suing over digitally created content, a plaintiff must give the publisher written notice, and removal within 10 days limits recovery to actual damages. It establishes an exclusive right to consent to use of one's identity and extends the Abuse of Personal Identity Act, as amended by 2025 SB 271, to distributing or trafficking in identity-replication tools, with exemptions. Effective May 6, 2026.
Law-enforcement agencies; use of certain technologies & interrogation practices; forensic lab.
Virginia's HB 1257 (2026 Regular Session) would direct the Department of Criminal Justice Services to set a framework for law-enforcement use of generative AI, machine learning systems and audiovisual surveillance technologies and for interrogations, with model policies and training. Continued to 2027 in committee on February 2, 2026.
To establish a commission to investigate AI in education
Not law. S 429 would have created a commission in the Massachusetts Executive Office of Education to study AI in schools and recommend regulations, limits, standards and safeguards. It would have surveyed how public and private school systems use generative AI and automated decision systems (for grading, scheduling, lesson planning and budgeting), examined bias, transparency, data protection, plagiarism and a possible watermark requirement, and produced guidelines for ethical use. On December 4, 2025 it was sent to a study order (S 2792), ending it for the 2025-2026 session.
Youth AI Privacy Act
The federal S. 4199 (Youth AI Privacy Act), as introduced, would require AI chatbot deployers that know a user is a minor to disclose that the user is talking to an AI, bar profiling, advertising and training on minors' data, and have the FTC write rules against compulsive-use features. The Commerce Committee ordered it reported with a substitute August 5, 2026; the substitute text is not yet published, and the bill has not passed either chamber.
AI Foundation Model Transparency Act of 2026
The federal H.R. 8094 (AI Foundation Model Transparency Act of 2026) would direct the FTC to issue regulations within 1 year of enactment requiring covered foundation model providers to submit information to the FTC and publish details on training data, testing and operations. Fully open-source models would be exempt. Referred to the House Energy and Commerce Committee March 26, 2026; it has not passed either house.
Artificial intelligence; age verification systems required for chatbots, safeguard protocols required, therapy chatbot requirements established, private right of action and enforcement provided for
Alabama's HB 324, as introduced, would require AI chatbot deployers to verify the age of every user, keep chatbots with human-like features away from minors (under 19) or offer minors a version without them, adopt protocols for users who express intent to harm themselves or others, and limit the data they collect and store. It would allow licensed psychiatrists to prescribe qualifying therapeutic chatbots to minors, with a private right of action and Attorney General enforcement. Referred to House Judiciary January 22, 2026; the bill died when the 2026 regular session ended without final passage.
California AI Transparency Act: system provenance data.
California's AB 2713 rewrites the large-online-platform duties of the California AI Transparency Act (Bus. & Prof. Code § 22757.3.1). From January 1, 2027, large online platforms must detect provenance data on content they distribute, show in their interface whether system provenance data or a digital signature identifies content as generated or substantially altered by a generative AI system or captured by a camera or other capture device, and let users inspect that data. They may not, to the extent technically feasible, knowingly strip it. Only data compliant or interoperable with widely adopted standards is covered. Signed September 30, 2026 (Chapter 856); effective January 1, 2027.
Consumer Protection and Product Liability - Chatbots
Maryland's SB 827 (2026, first reader text, cross-filed with HB 1250) would have regulated operators and developers of generative AI chatbots made available to Maryland users: affirmative consent for training on input data, a ban on using chat logs for advertising, a 10-year cap on chat log retention, protections for users under 13, hourly not-human warnings, monthly safety testing, and product liability with possible strict liability. It had a Senate Finance hearing set for March 12, 2026 and died when the 2026 session ended.
State management: other; artificial intelligence pilot program; establish for state departments and agencies. Creates new act.
Michigan's HB 5899 (2026), by Rep. Jaime Greene, would create the Artificial Intelligence Pilot Program Act. The Department of Technology, Management, and Budget (DTMB) would run a pilot, beginning no later than January 1, 2027, letting approved state employees, agencies and departments use generative AI, guided by a three-member AI governing board appointed by the governor. Introduced April 23, 2026; the Communications and Technology Committee reported it for referral to the Rules Committee May 19, 2026. It has not passed either chamber.
AI WORK TRANSITION COMMISSION
Illinois's HB 5760 (104th General Assembly) would create the Commission on Artificial Intelligence and Workforce Transition Act. The Commission would direct State agencies to assess the actual and projected impact of generative and frontier AI systems on Illinois industries, occupations, regions and workers, and make recommendations on State and federal workforce, education, apprenticeship, unemployment insurance and income support programs. The Act would be repealed January 1, 2034. Last action: referred to the House Rules Committee May 5, 2026.
Artificial Intelligence Bill of Rights
Florida's SB 2-D (2026 special session D), the Artificial Intelligence Bill of Rights, would have required companion chatbot platforms to get parental consent before a minor holds an account, required bots to tell users hourly that they are not talking to a human, barred AI companies from selling users' personal information unless deidentified, restricted AI tools in schools, and required AI vendors to government to attest they are not owned, controlled or based in a foreign country of concern. It passed the Senate 37-1 on April 28, 2026, was referred in the House to the Information Technology Budget & Policy Subcommittee, and died there on April 29, 2026. It did not become law.
AN ACT relating to protection of information and declaring an emergency.
Kentucky's SB 4 (2025) governs state government's use of AI. The Commonwealth Office of Technology must create an Artificial Intelligence Governance Committee, set standards for agencies' use of generative and high-risk AI, keep a registry of those systems and report yearly. Agencies must disclose AI use to the public, explain AI-driven decisions and offer appeals of consequential decisions. A candidate whose appearance, action or speech is altered with synthetic media in an electioneering communication may sue the sponsor to require a clear disclosure. Signed March 24, 2025 (Acts Ch. 66); emergency act, effective on signing.
AN ACT relating to campaign practices; requiring certain communications relating to an election that include synthetic media to contain a disclosure; and providing other matters properly relating thereto.
Nevada AB 73 adds NRS 294A.3493. A campaign communication paid for by a person, candidate or committee that supports or opposes a candidate or party and includes synthetic media (an image, audio or video of a person intentionally manipulated with AI, generative AI or GAN techniques so it looks real but gives a fundamentally different impression) must clearly state "This (image/video/audio) has been manipulated." Video must show it throughout; audio must say it at the start, end and every 2 minutes. A depicted candidate can sue for an injunction. Broadcasters, platforms and satire/parody are exempt. Effective Jan. 1, 2026.
Prohibits advertising generative artificial intelligence as able to practice regulated profession or occupation.
New Jersey's S4088 would bar developers and deployers of generative AI in the State from advertising or representing to the public that the AI is able to practice a profession or occupation regulated under Title 45, by the Judicial Branch or by the Administrative Office of the Courts. A violation would be an unlawful practice under the consumer fraud act. Introduced May 4, 2026.
AN ACT to amend Tennessee Code Annotated, Title 39 and Title 40, relative to the sexual exploitation of children.
Tennessee HB 2163 (Public Chapter 911) brings AI-created imagery within the state's sexual exploitation of children offenses. The definition of 'material' in TCA 39-17-1002 now includes any computer image or computer-generated image, including an image created, adapted or modified by artificial intelligence. The act defines artificial intelligence, including generative AI that creates new text, images, audio or video when prompted. Effective July 1, 2024.
Law Enforcement Usage of Artificial Intelligence
Utah SB 180 (2025) requires each law enforcement agency to have a policy on employees' use of generative AI, covering approved tools, permitted tasks, review of AI output and discipline for violations. A police report or other law enforcement record created wholly or partly with generative AI must carry a disclaimer that it contains AI-generated content, and the author must certify that they read and reviewed it for accuracy. The act defines artificial intelligence and generative AI. Effective May 7, 2025.
Identifying Outputs of Generative Adversarial Networks Act (IOGAN Act)
The IOGAN Act directs the National Science Foundation to support research on manipulated or synthesized content and information authenticity, including digital forensic tools, watermarking of generated media, and public awareness of deepfakes (sec. 3). It directs NIST to support measurement and standards research to speed development of tools that examine the outputs of generative adversarial networks (sec. 4). NSF and NIST must jointly report to Congress within one year on the feasibility of public-private partnerships to detect manipulated or synthesized content (sec. 5). Section 6 defines generative adversarial network.
Requires policing agencies to conduct an inventory of, and develop a publicly-available policy for, any artificial intelligence used to aid criminal investigations.
New York's S 10425 would add Executive Law § 837-ww requiring policing agencies to publish a disclosure for any "covered AI" used to aid criminal investigations, including generative AI used to write police reports, at least 90 days before using a new product. Agencies would have to disclose AI use in filed reports, keep audit trails, and adopt an Attorney General model policy. Referred to Codes May 15, 2026; it has not passed either chamber.
State-by-State Regulation of Generative AI
Artificial intelligence regulation is advancing at the state level. This page lists the 314 published bills we track on Generative AI. State attorneys general, privacy regulators, and legislatures are developing a patchwork of laws that impose different obligations on businesses deploying AI systems related to this topic. Compliance teams and legal counsel must track both enacted laws and the pipeline of pending bills that could shift requirements at any time.
Obligations differ from bill to bill. Depending on the law, they can include consumer disclosure obligations, prohibition on specific high-risk uses, mandatory algorithmic impact assessments, bias audits, rights to explanation, and human override requirements. Penalties for non-compliance vary by state — use our Penalty Tracker to review enforcement provisions. For upcoming effective dates, see the Deadlines page.
To understand how Generative AI laws intersect with specific industries, browse our Healthcare, Employment, Insurance, and Financial Services industry trackers. Compare how different state laws approach this topic using our Bill Comparator.