AI bias audit laws require organizations deploying automated decision systems to conduct independent audits evaluating whether their AI produces discriminatory outcomes. New York City's Local Law 144 is the best-known bias audit mandate: it requires an independent audit and public posting of results before an employer uses an automated employment decision tool. State proposals extend similar testing ideas to lending, housing, insurance and other high-stakes domains.
Key regulatory themes include mandatory annual or biennial audits, qualified auditor requirements, disparate impact analysis across protected classes, public reporting obligations, and penalties for non-compliance. Several states are also requiring pre-deployment impact assessments that overlap with bias audit requirements.
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.
Enacts the New York privacy act to require companies to disclose their methods of de-identifying personal information, to place special safeguards around data sharing and to allow consumers to obtain the names of all entities with whom their information is shared.
New York's A 8158 (the "New York privacy act") would add General Business Law Article 42-A (§§ 1200-1207). Among other consumer rights, people could opt out of "profiling in furtherance of decisions that produce legal or similarly significant effects" (such as lending, housing, insurance, employment and health care), and controllers would have to run data protection assessments for profiling that presents a reasonably foreseeable risk of unfair treatment or unlawful disparate impact. Only the Attorney General could enforce it, with penalties of up to $20,000 per violation. Introduced May 2, 2025 and referred to Consumer Affairs and Protection (again January 7, 2026); not passed.
Establishes criteria for the use of automated employment decision tools; provides for enforcement for violations of such criteria.
New York's S 4394-A would add Labor Law Article 20-D (§§ 750-757) for automated employment decision tools, AI systems specifically developed, marketed or modified to be the controlling factor in a decision to hire, promote or terminate. Deployers may not use them in a way that violates Executive Law Article 15; deployers and developers must perform impact assessments within one year and annually, keep governance programs and give notice to individuals. The Attorney General enforces after notice and a 45-day cure period. Referred to Labor January 7, 2026.
Relates to the use of automated decision tools by landlords for making housing decisions; sets conditions and rules for use of such tools.
New York's S 7735-A (2023-2024) would have added Real Property Law § 227-g regulating landlords' automated decision tools for housing decisions. Landlords would need an independent auditor's disparate impact analysis at least annually, post a summary on their website, and notify applicants at least 24 hours before use, including reasons for a denial. The Attorney General could investigate and sue. Amended April 2, 2024 and left in committee; it died when the 2023-2024 session ended.
Consumer Protection - High-Risk Artificial Intelligence - Developer and Deployer Requirements
Maryland's SB 936 (2025) would have added Commercial Law sections 14-47A-01 to 14-47A-08 requiring developers and deployers of high-risk AI systems used for consequential decisions to use reasonable care against algorithmic discrimination starting February 1, 2026, with impact assessments, consumer notices and appeals. The Attorney General would enforce, with fines up to $1,000 per violation ($1,000 to $10,000 if willful). A Senate Finance hearing was set for February 27, 2025; the bill went no further and died when the session ended.
Consumer Protection - Consumer Reporting Agencies - Use of Algorithmic Systems
Maryland's HB 1399 (2026, first reader text) would have added Commercial Law section 14-1228 requiring consumer reporting agencies that use algorithmic systems to assemble or evaluate consumer information to explain evaluations in plain language, keep a public registry of algorithms, meet error-rate targets, get quarterly independent bias audits, have humans review automated evaluations within 24 hours, and notify consumers when AI is used. It had a House Economic Matters hearing set for March 10, 2026 and died when the 2026 session ended.
Relating To Artificial Intelligence.
Hawaii's SB 2967 (2026) would have added a part to HRS chapter 480 regulating businesses ('deployers') that use AI systems or automated decision tools with consumers. They would have to disclose AI interactions, give notice and explanation, correction, appeal and human review for AI-influenced consequential decisions, keep risk programs, do annual impact assessments and report incidents within 90 days. A violation would be an unfair or deceptive act. The Senate committee on LBT deferred it February 11, 2026; it died.
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.
Ban on employer use of automated decision systems.
Not law. HB 1421 would have barred Indiana employers with 11 or more workers, and state and local government employers, from relying solely on an automated decision system for hiring, firing, pay, scheduling or other job decisions. Using such a system's output would have required bias testing before use and independent testing every year, advance notice to workers and applicants, human review, plain-language documentation within 7 days, and a right to dispute and appeal to a person. Workers could opt out of algorithmic management. It was referred to committee on January 8, 2026 and got no further action before the 2026 session adjourned.
Enacts the legislative oversight of automated decision-making in government act (LOADinG Act) to regulate the use of automated decision-making systems and artificial intelligence techniques by state agencies.
New York's S 7543-B, the LOADinG Act (Chapter 674 of 2024), as signed would have barred state agencies from using automated decision-making systems for public benefits or decisions affecting rights or safety without meaningful human review, and required impact assessments. Before those rules took effect, Chapter 96 of 2025 replaced them: each state agency using automated employment decision-making tools must post a list of them by December 30 each year, starting 2025; the Office of Information Technology Services keeps a public AI inventory; and, until July 1, 2028, AI use may not displace state employees. Signed December 21, 2024; effective July 1, 2025.
Regulates automated decision-making by government agencies; requires agencies to conduct impact assessments; requires disclosure of automated decision-making tools utilized by governmental agencies.
New York's S 7599-C (Chapter 711 of 2025) added State Technology Law Article 5, under which agencies would have listed automated decision-making tools and done bias-tested impact assessments from December 19, 2026. Chapter 86 of 2026 repealed Article 5, which had not yet taken effect, and the § 103-f state tool inventory. Now in force: public employers' use of AI systems may not displace employees or impair collective bargaining, and (added by Chapter 86) local governments, school districts, BOCES, SUNY, CUNY and community colleges must, where feasible, list automated employment decision tools yearly from December 30, 2026. Signed December 19, 2025; expires July 1, 2028.
Provides for the use of automated employment decision-making tools and artificial intelligence systems by a county, city, town, village, school district, board of cooperative educational services, county vocational education and extension board, district corporation, the state university of New…
New York's S 8831 (Chapter 86 of 2026) is the chapter amendment to Chapter 711 of 2025. It repealed Chapter 711's State Technology Law Article 5 (impact assessments, bias testing and public lists of government automated decision-making tools, set to start December 19, 2026), its state tool inventory and its report to the Legislature. Instead, local governments, school districts, BOCES, SUNY, CUNY and community colleges must, to the extent feasible, post yearly lists of automated employment decision tools, first by December 30, 2026, and public employers may not use AI systems to displace employees. Signed February 13, 2026, with effect from December 19, 2025; the rules expire July 1, 2028.
Adopt the Artificial Intelligence Consumer Protection Act
Not law. Nebraska LB642, introduced by Senator Bostar, would have enacted the Artificial Intelligence Consumer Protection Act. Developers and deployers of high-risk AI systems that make consequential decisions without human review would have had to use reasonable care against algorithmic discrimination, with impact assessments, consumer notices and appeals, and Attorney General enforcement after a 90-day cure period. Indefinitely postponed April 17, 2026.
ALGORITHMIC IMPACT ASSESSMENTS
Illinois's HB 5322 (103rd General Assembly) would have created the Illinois Commercial Algorithmic Impact Assessments Act. By January 1, 2026, and annually after, deployers and developers of an automated decision tool would have to complete and document an impact assessment, and developers would have to give deployers documentation on intended uses and limitations. It was not passed before the 103rd General Assembly ended (Session Sine Die, January 7, 2025).
Creates standards for independent bias auditing of automated employment decision tools.
New Jersey's A3855, as introduced, would bar an employer or employment agency from using an automated employment decision tool (AEDT) if more than one year has passed since its last independent bias audit, require audits by sex, race/ethnicity and intersectional categories, and require posted audit results and candidate notice at least 10 business days before use; it sets no penalties. Introduced February 22, 2024. The Assembly Science, Innovation and Technology Committee reported an Assembly Committee Substitute on May 16, 2024, whose text is not described here. A bill of the Legislature's 2024-2025 session, which ended January 13, 2026.
Virginia Human Rights Act; equal credit opportunities, Virginia Fair Housing Law, nondiscrimination.
Virginia's HB 999 (2026 Regular Session) would have made it an unlawful discriminatory practice to use an automated decision system in decisions under the Human Rights Act, equal credit law or Fair Housing Law that results in discrimination. The House General Laws substitute focused on credit. It failed: left in House Appropriations Committee on February 18, 2026.
Establishes the New York workforce stabilization act; requires certain businesses to conduct artificial intelligence impact assessments on the application and use of such artificial intelligence and to submit such impact assessments to the department of labor prior to the implementation of the…
New York's S 1854-A, the New York workforce stabilization act, would add Labor Law § 201-j barring employers with more than 100 employees from using AI without a prior impact assessment submitted to the Department of Labor at least 30 days before implementation. It would also add Tax Law § 186-h, a 2% surcharge on business income base for corporations displacing workers through automation or using AI for data mining. Referred to Labor January 7, 2026; it has not passed either chamber.
Relating to regulation of the use of artificial intelligence systems in this state; providing civil penalties.
Texas's HB 149, the Texas Responsible Artificial Intelligence Governance Act, bars developing or deploying AI intended to incite self-harm, harm to others or crime, with intent to unlawfully discriminate against a protected class (disparate impact alone is not enough), or solely to infringe U.S. constitutional rights, and making or distributing AI solely to produce or spread child pornography or unlawful sexual deep fakes. Government entities may not use AI for harmful social scoring or rights-infringing biometric ID. Agencies must tell consumers they are dealing with AI; health care providers must tell patients. AG enforcement; 60-day cure. Signed June 22, 2025; effective January 1, 2026.
Safe and Secure Innovation for Frontier Artificial Intelligence Models Act.
Vetoed. California's SB 1047 (2024) would have required developers of covered AI models (trained with more than 10^26 operations at a compute cost over $100 million) to adopt safety and security protocols and full-shutdown capability before training, assess and guard against critical harms before release, hire annual third-party auditors from 2026, and report AI safety incidents to the Attorney General within 72 hours. The Attorney General could have sued for penalties. Governor Newsom vetoed it September 29, 2024, and the Legislature did not override the veto.
Consumer Protection - Online Products and Services - Data of Children (Maryland Kids Code)
Maryland's HB 603 (Kids Code) covers for-profit firms doing business in Maryland that have over $25M revenue, buy, receive, sell or share data of 50,000+ consumers, households or devices yearly, or get at least half their revenue from data sales. For online products reasonably likely to be accessed by under-18s they must do impact assessments (existing ones by April 1, 2026) and set high-privacy defaults. AI-related: no default profiling of children without safeguards and need or a compelling reason; assessments weigh algorithm harms. Fines up to $7,500 per child per violation. Signed May 9, 2024 (Chapter 461; same as SB 571); effective October 1, 2024. Not enjoined; NetChoice suit pending.
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.
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 Concerning Artificial Intelligence, Automated Decision-making And Personal Data Privacy.
Connecticut's Public Act 23-16 governs AI in state government. The Department of Administrative Services (for all state agencies) and the Judicial Department must inventory the AI systems in use by December 31, 2023 and yearly after, and post them. OPM and the Judicial Branch set policies by February 1, 2024 for buying, using and assessing AI; from then, agencies may not deploy an AI system without an impact assessment showing it will not cause unlawful discrimination or disparate impact. It formed a working group on AI best practices and an AI bill of rights, and from October 1, 2023 state contracts must require compliance with the data privacy act. Signed June 7, 2023.
Establishes the New York workforce stabilization act; requires certain businesses to conduct artificial intelligence impact assessments on the application and use of such artificial intelligence and to submit such impact assessments to the department of labor prior to the implementation of the…
New York's S 9401 (2023-2024), the Workforce Stabilization Act, would have barred employers (in-state businesses with more than 100 employees) from using AI unless they first conducted an impact assessment, repeated at least every two years and submitted to the Department of Labor at least thirty days before implementation. It would also have imposed a 2% surcharge on corporations that terminate or substantially reduce hours for 15 or more employees because of AI, and on AI data mining. Referred to Labor May 15, 2024; it died at the end of the 2023-2024 session.
Enacts the New York privacy act to require companies to disclose their methods of de-identifying personal information, to place special safeguards around data sharing and to allow consumers to obtain the names of all entities with whom their information is shared.
New York's A 974-A, the "New York data protection act," would add General Business Law Article 42-A letting consumers opt out of profiling (automated processing of personal data to evaluate or predict personal traits) in furtherance of decisions that produce legal or similarly significant effects, and requiring a data protection assessment for profiling with a reasonably foreseeable risk of unfair or deceptive treatment of, or unlawful disparate impact on, consumers. It would also add targeted-advertising and data-sale opt-outs and other rights, with Attorney General civil penalties of up to $20,000 per violation. Amended May 29, 2026 and recommitted to Codes; not passed by either chamber.
Creates standards for independent bias auditing of automated employment decision tools.
New Jersey's A1021 would bar an employer or employment agency from using an automated employment decision tool if more than one year has passed since its most recent independent bias audit. Audits would have to calculate selection or scoring rates and impact ratios across categories such as sex, race, age and disability. Pre-filed for introduction in the 2026 session.
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).
Labor and Employment – Automated Employment Decision Tools – Prohibition
Maryland's SB 957 (2024) would have barred employers from using automated employment decision tools to screen applicants or decide terms of employment unless the tool passed a yearly impact assessment finding no high-risk action, and required employers to notify affected applicants within 30 days of use. Civil penalties would start at up to $500 per violation. A Senate Finance hearing was set for March 14, 2024; it went no further and died when the session ended.
Consumer Protection - Consumer Reporting Agencies - Use of Algorithmic Systems
Maryland's HB 1477 (2025) would have added Commercial Law section 14-1228 for consumer reporting agencies that use algorithmic systems to assemble or evaluate consumer information. They would have had to give plain-language explanations, keep a public registry of algorithms, meet error-rate and accuracy thresholds, get quarterly third-party bias audits, and have all automated evaluations reviewed by a human within 24 hours. It was introduced February 7, 2025, its March 4 hearing was canceled, and it died when the session ended.
CONSUMER DATA PRIVACY
Illinois's SB 3220 would create the Illinois Consumer Data Privacy Act, giving consumers rights to access, correct, delete and port their personal data and to opt out of targeted advertising, data sales and profiling used for decisions with legal or similarly significant effects. Controllers would have to run data protection impact assessments, including for risky profiling. The Attorney General would enforce it. Filed February 2, 2026; re-referred to Assignments May 22, 2026; it has not passed either house.
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.
AN ACT RELATING TO LABOR AND LABOR RELATIONS -- ARTIFICIAL INTELLIGENCE USE AND FAIR EMPLOYMENT PRACTICES
Rhode Island's H 7767 would add a chapter to Title 28 limiting how employers use electronic monitoring and automated decision systems on workers. Employers could monitor only for listed purposes, with written notice, an independent impact assessment, and human review of decisions based on monitoring data; monitoring tools using facial recognition or gait, voice analysis or emotion recognition would be barred. Introduced February 12, 2026; House Labor recommended it be held for further study April 15, 2026. It has not passed either chamber.
AUTOMATED DECISION TOOLS
Illinois's SB 2203 (104th General Assembly) would create the Preventing Algorithmic Discrimination Act. On or before January 1, 2027, and annually after, a deployer of an automated decision tool would have to perform an impact assessment, notify people subject to consequential decisions, and keep a governance program. Re-referred to Assignments on April 11, 2025.
AI-Meaningful Human Review
Illinois's HB 3567 (104th General Assembly) would create the Meaningful Human Review of Artificial Intelligence Act. A State agency, or an entity acting for it, could not use an automated decision-making system for specified agency functions without continuous meaningful human review, and the agency would have to complete an impact assessment submitted to the Governor and General Assembly. Re-referred to Rules on March 27, 2026.
AI-MEANINGFUL HUMAN REVIEW
Illinois's HB 3720 (104th General Assembly) would create the Meaningful Human Review of Artificial Intelligence Act. A State agency, or an entity acting for one, would be prohibited from using an automated decision-making system for specified agency functions without continuous meaningful human review, and agencies seeking to use such a system would have to perform impact assessments and submit them to the Governor and General Assembly. Re-referred to the House Rules Committee on March 21, 2025.
CONSUMER DATA PRIVACY ACT
Illinois's SB 3548 (104th General Assembly) would create the Consumer Data Privacy Act, giving consumers rights to confirm whether personal data is processed, correct and delete it, obtain a copy, and opt out of targeted advertising, data sales or profiling. It would apply to persons controlling or processing data of at least 100,000 consumers, with Attorney General enforcement. Last action: re-referred to Assignments May 22, 2026.
AGE-APPROPRIATE DESIGN CODE
Illinois' SB 50 is mainly a children's online privacy and design-code bill. As introduced, it would require covered entities (for-profit businesses over set revenue or data thresholds) to complete data protection impact assessments for services reasonably likely to be accessed by children, including whether algorithms and targeted advertising systems could harm children, and would bar profiling a child by default. Introduced January 13, 2025; re-referred to Assignments April 11, 2025 and still pending.
AGE-APPROPRIATE DESIGN CODE
Illinois' SB 3241 is mainly a children's online privacy and design-code bill. As introduced, it would require covered entities (for-profit businesses over set revenue or data thresholds) to complete data protection impact assessments for services reasonably likely to be accessed by children, including whether algorithms and targeted advertising systems could harm children, and would bar profiling a child by default. Introduced February 2, 2026; re-referred to Assignments May 22, 2026.
AGE-APPROPRIATE DESIGN CODE
Illinois' SB 51 is mainly a children's online privacy and design-code bill. As introduced, it would require a business whose online service, product, or feature is likely to be accessed by people under 18 to complete a data protection impact assessment, which must address whether the service's algorithms and targeted advertising systems could harm children, and would bar profiling a child by default. Attorney General enforcement only. Introduced January 13, 2025; re-referred to Assignments April 11, 2025 and still pending.
Establishes Artificial Intelligence Ethics Board in DOLWD.
A 2478 would create a seven-member Artificial Intelligence Ethics Board in New Jersey's Department of Labor and Workforce Development to oversee the sale and use of AI technology, with a focus on bias prevention, privacy, transparency and accountability. Anyone selling or offering an AI product or service in the state would first need an independent AI impact assessment, done within the prior year and meeting the board's standards on privacy and job displacement. The board would also recommend protections for workers displaced by AI, such as a month's notice, retraining and severance. Pending in the Assembly Science, Innovation and Technology Committee.
Establishes the New York workforce stabilization act; requires certain businesses to conduct artificial intelligence impact assessments on the application and use of such artificial intelligence and to submit such impact assessments to the department of labor prior to the implementation of the…
New York's A 5429-A (the "New York workforce stabilization act") would add Labor Law § 201-j barring employers (in-state businesses with more than 100 employees) from using AI without an impact assessment, repeated every two years and submitted to the department 30 days before implementation, and add Tax Law § 186-h imposing a surcharge of two percent of business income base on corporations that displace employees through AI above tiered thresholds or use AI for data mining. Section 2 takes effect immediately; the tax section on January 1, 2026. Referred to Labor January 7, 2026; it has not passed either chamber.
Office of Artificial Intelligence Policy Amendments
Utah's HB 320 revises the existing Office of Artificial Intelligence Policy and its AI Learning Laboratory (Title 13, Chapter 72). Besides regulatory mitigation agreements, which let a participant use AI despite a law or rule that might impede it, the office may sign joint interpretation agreements clarifying how a state law or rule applies to an AI technology. Agreements must set consumer disclosures and audit reporting; the office must audit participants regularly and may end an agreement at any time. Terms run up to 12 months with up to two extensions. Signed March 18, 2026; effective May 6, 2026.
Restricts the use by an employer or an employment agency of electronic monitoring or an automated employment decision tool to screen a candidate or employee for an employment decision unless such tool has been the subject of an impact assessment within the last year; requires notice to employment…
New York's A 3779-A (the "boundaries on technology act") would add Labor Law Article 35-A: employers with 100 or more employees could not use an automated employment decision tool without an impact assessment by an impartial party within the prior year and yearly after; all employers using such tools would owe notice, meaningful human oversight and data access rights. The Labor Commissioner could assess penalties under Labor Law § 218 and the Attorney General could sue. Would take effect on the 180th day after becoming law. On the Assembly floor calendar since May 29, 2026; it has not passed either chamber.
Relates to the use of automated decision tools by landlords for making housing decisions; sets conditions and rules for use of such tools.
New York's A 3125-A would add Real Property Law § 227-g making it unlawful for a landlord to use an automated housing decision making tool to screen applicants unless a disparate impact analysis by an independent auditor is done at least annually and a summary is posted on the landlord's website before use. Applicants would have to get notice at least 24 hours before use and may request an alternative process. The Attorney General could investigate and sue. Would take effect immediately. Ordered to third reading (calendar 103) in the Assembly January 7, 2026; it has not passed either chamber.
Establishes criteria for the use of automated employment decision tools; provides for enforcement for violations of such criteria.
New York's A 3914 would add Labor Law § 203-g making it unlawful for an employer to use an automated employment decision tool to screen candidates unless a disparate impact analysis is done at least annually, a summary and the tool's distribution date are posted on the employer's website before use, and the summary is provided annually to the Department. The Attorney General or Commissioner could investigate and sue. Would take effect immediately. Referred to the Labor Committee January 30, 2025 and January 7, 2026; it has not passed either chamber.
Enacts the NY privacy act to require companies to disclose their methods of de-identifying personal information, to place special safeguards around data sharing and to allow consumers to obtain the names of all entities with whom their information is shared.
New York's A 4947 (the "New York privacy act") would add General Business Law Article 42-A. For solely automated decisions that materially contribute to denials of lending, housing, insurance, health care or similar services, controllers would have to disclose the automation, offer an appeal with meaningful human review and run annual impact assessments by an external, independent auditor or researcher. Consumers would also get opt-in consent, access, correction and deletion rights. AG penalties up to $15,000 per violation; private suits. Most sections would take effect two years after enactment. Referred to committee February 10, 2025 and January 7, 2026; not passed by either chamber.
STATE AGENCIES-AI SYSTEMS
Illinois's HB 4836 (103rd General Assembly) would have added Section 5-735 to the Civil Administrative Code so that State agency and State-funded AI systems must follow the NIST AI Risk Management Framework. Agencies and entities deploying State-funded AI would have to submit NIST-based algorithmic impact assessments to the General Assembly, the Auditor General and the Department of Innovation and Technology. It was not passed before the 103rd General Assembly ended (Session Sine Die, January 7, 2025).
Requires real estate brokers and online housing platforms using virtual agents and AI tools to conduct a disparate impact analysis to assess the actual impact of any such virtual agent; requires such individuals to submit a summary of the most recent disparate impact analysis to the attorney…
New York's A 9028 would amend Real Property Law § 440 and add § 442-m. At least once a year, brokers or online housing platforms using virtual agents for property searches, and platforms using AI tools, would have to have an independent auditor conduct a disparate impact analysis and submit a summary to the Attorney General's office, test for and correct discriminatory results, and (for AI tools) follow ad and captioning safeguards and report publicly. Would take effect immediately. Referred to Judiciary September 5, 2025 (and January 7, 2026); it has not passed either chamber.
AUTOMATED DECISION TOOLS
Illinois's HB 5116 (103rd General Assembly) would have created the Automated Decision Tools Act. On or before January 1, 2026, and annually after, a deployer of an automated decision tool would have to perform an impact assessment, notify people subject to a consequential decision, and keep a governance program to manage algorithmic discrimination risk. The Attorney General could sue. Died at Session Sine Die on January 7, 2025.
Enacts the "New York artificial intelligence consumer protection act", in relation to preventing the use of artificial intelligence algorithms to discriminate against protected classes.
New York's S 1962, the New York artificial intelligence consumer protection act, would add General Business Law Article 45-A (sections 1550 to 1556). From January 1, 2027, developers and deployers of high-risk AI decision systems that make or substantially factor in consequential decisions (education, employment, lending, government services, health care, housing, insurance, legal services) would have to use reasonable care against algorithmic discrimination, with impact assessments, notices and appeals. The Attorney General would have exclusive enforcement. Referred to Internet and Technology January 7, 2026; it has not passed either chamber.
Relates to the use of automated lending decision-making tools by banks for the purposes of making lending decisions; allows loan applicants to consent to or opt out of such use.
New York's S 8115-C would add Banking Law section 103-a on automated lending decision-making tools used by banking organizations, licensed foreign and interstate branches and licensed lenders (national and federal institutions are excluded). Covered entities must get an annual independent impact assessment, post the final report on their website, keep records for seven years, report discriminatory outcomes to the department within thirty days, and give applicants notice at least 24 hours before use. Reported from Banks and committed to Internet and Technology May 12, 2026; it has not passed either chamber.
Restricts the use by an employer or an employment agency of electronic monitoring or an automated employment decision tool to screen a candidate or employee for an employment decision unless such tool has been the subject of an impact assessment within the last year; requires notice to employment…
New York's S 185-A would add Labor Law Article 37 (sections 1050 to 1057) limiting employer electronic monitoring and automated employment decision tools (AEDTs). Employers could not use an AEDT for an employment decision without an independent impact assessment within the prior year, repeated annually, or without 10 business days' notice; they may not rely primarily on tool output and must stop use if disparate impact is found. Monitoring must be strictly necessary and noticed in advance. It adds an employee private action with liquidated damages. Amended April 17, 2025; it has not passed either chamber.
Relates to the use of automated decision tools by landlords for making housing decisions; sets conditions and rules for use of such tools.
New York's S 6471-A would add Real Property Law § 227-g making it unlawful for a landlord to use an automated housing decision making tool to screen applicants unless an independent disparate impact analysis is done at least annually. A summary of the latest analysis must be posted on the landlord's website and on listings, and applicants must get notice at least 24 hours before use. The Attorney General could sue to compel compliance. Referred to Investigations and Government Operations January 7, 2026; it has not passed either chamber.
Relates to the use of automated lending decision-making tools by banks for the purposes of making lending decisions; allows loan applicants to consent to or opt out of such use.
New York's A 773-C would add Banking Law § 103-a requiring covered banking organizations and licensed lenders that use automated lending decision-making tools to commission an annual independent impact assessment and post a summary report, report discriminatory or biased findings to the Department within thirty days, and give applicants notice at least 24 hours before use and, to the extent practicable, a denial reason within 24 hours. It contains no consent or opt-out right. Reported and advanced to third reading (calendar 393) April 30, 2026; it has not passed either chamber.
Artificial intelligence, consumer protection
South Carolina's S 963 would enact the Consumer Protections in Interactions with Artificial Intelligence Systems Act (Chapter 31 of Title 37). Developers and deployers of high-risk AI systems used in consequential decisions would owe a duty of reasonable care against algorithmic discrimination, with impact assessments, consumer notices and appeal rights. The Attorney General would have exclusive enforcement. Introduced February 26, 2026 and referred to the Senate Banking and Insurance Committee; it has not passed either chamber.
Labor: fair employment practices; use of electronic monitoring or automated decisions tools by an employer; prohibit except for certain purposes. Creates new act.
Michigan's HB 5579, the 'responsible artificial intelligence security for employees act', would bar employers from using an automated decisions tool to make an employment-related decision, except to screen large volumes of applications or assess job skills. It would also limit electronic monitoring, require independent impact assessments, and allow civil fines of up to $500. Introduced February 24, 2026 and referred to the House Committee on Economic Competitiveness; it has not passed either chamber.
Regulating high-risk artificial intelligence system development, deployment, and use.
Washington's SB 6120 (2026 session), the Senate companion to HB 2157, would have required developers and deployers of high-risk AI systems, those intended to make or be a substantial factor in consequential decisions such as employment, lending, housing, insurance, health care and education, to use reasonable care against algorithmic discrimination. It included developer documentation, deployer risk programs and impact assessments, consumer notices, and private lawsuits for injunctions. It had a first reading on January 14, 2026 and died at adjournment.
Preventing Algorithmic Collusion Act of 2025
The federal S 232 (Preventing Algorithmic Collusion Act of 2025) would make it unlawful to use or distribute a pricing algorithm that uses or was trained with nonpublic competitor data, with federal civil penalties. It would also require audit reports on request and disclosures of algorithmic pricing by companies with $5 million or more in revenue. Read twice and referred to the Senate Judiciary Committee January 23, 2025; it has not passed either house.
Establishes consumers' foundational data privacy rights; creates oversight mechanisms; establishes enforcement mechanisms; establishes the privacy and security victims relief fund.
New York's A5827 (2025-26) would add a new Article 46, the "American Data Privacy and Protection Act", to the General Business Law, covering data minimization, a duty of loyalty, consumer rights to access, correct, delete and port covered data, and consent and opt-out rules for transfers and targeted advertising; its title also establishes a privacy and security victims relief fund. A large data holder using a covered algorithm in a way that poses a consequential risk of harm would have to conduct an impact assessment of it. Referred to the Assembly Science and Technology Committee February 20, 2025 and again January 7, 2026; not enacted.
STATE GOVT AI ACT
Illinois' SB1366 would create the State Government AI Act. The Department of Innovation and Technology would have to adopt rules before January 1, 2028 on the development, procurement, deployment, use and assessment of AI by State agencies, and from January 1, 2028 agencies could not deploy or use AI unless those rules permit it. Filed January 29, 2025 and referred to the Senate Assignments Committee; it has not passed either house.
Regulates use of automated employment decision tools in employment decisions to minimize discrimination in employment.
New Jersey's A2726 would regulate automated employment decision tools. A tool could not be sold, developed, deployed, used or offered in New Jersey unless it had a bias audit in the past year; employers would have to notify candidates at least 10 days before use and post audit summaries. Civil penalties would be up to $1,500 for a developer's first violation and up to $500 per violation for employers. Pre-filed for introduction in the 2026 session.
Creates "New Jersey Responsible AI Advancement and Workforce Protection Act."
New Jersey's S1840 would create the New Jersey Responsible AI Advancement and Workforce Protection Act, funding worker retraining through an AI Horizon Fund fed by a 5% assessment on AI infrastructure entities' gross AI revenue. It would require environmental assessments and community benefit agreements for AI computing facilities and impact assessments for high-risk AI systems.
Further providing for definitions; providing for use of automated employment decision tool; and further providing for civil penalties.
Pennsylvania's HB 594 (2025-26) would amend the Pennsylvania Human Relations Act to require an employer or employment agency using an automated employment decision tool to notify and get consent from individuals before the interview and to use only tools with a bias audit from within the past year, with the results summary posted online. It was referred to the House Labor & Industry Committee on February 12, 2025.
Protecting consumers in interactions with artificial intelligence systems
Massachusetts's H 97 would add Chapter 93M requiring developers and deployers of high-risk AI systems (those making or substantially influencing decisions on education, employment, finance, government services, health care, housing, insurance or legal services) to use reasonable care against algorithmic discrimination, with documentation, risk programs, impact assessments and consumer notices, and requiring disclosure when consumers interact with AI. Reported favorably and referred to House Ways and Means December 24, 2025; it has not passed either house.
Enacts the "digital fairness act"; requires any entity that conducts business in New York and maintains the personal information of 500 or more individuals to provide meaningful notice about their use of personal information; establishes unlawful discriminatory practices relating to targeted…
New York's S 4276 (the digital fairness act) would require businesses holding personal information of 500 or more people to give short-form privacy notices, get opt-in consent, and bar discriminatory targeted advertising. It would also set rules for government automated decision systems: use policies, notice, human review and impact assessments. Referred to the Senate Internet and Technology Committee February 3, 2025, and again January 7, 2026; it has not passed either house.
Enacts the New York Artificial Intelligence Civil Rights Act establishing protections for individual rights with respect to computational algorithms; establishes protections against the use of algorithms involved in consequential decisions, such as those that impact people's rights, civil…
New York's A 9654 (the New York Artificial Intelligence Civil Rights Act) would bar developers and deployers from using covered algorithms in consequential actions (such as employment, housing, credit, health care, insurance, education, criminal justice and government services) in ways that discriminate or have a disparate impact on protected characteristics. It would require independent audits where harm is plausible. Referred to the Assembly Science and Technology Committee January 21, 2026; it has not passed either house.
AN ACT RELATING TO LABOR AND LABOR RELATIONS -- ARTIFICIAL INTELLIGENCE USE AND FAIR EMPLOYMENT PRACTICES
Rhode Island SB 2499 (2026), as introduced on February 6, 2026, would add a chapter on electronic monitoring and automated decision systems in employment. It would make it unlawful for an employer to use an electronic monitoring tool to collect employee information unless stated conditions are met, bar tools that incorporate facial recognition, and require impact assessments. Employers would have to keep records for five years; penalties would be those in title 28. The text says it takes effect upon passage.
Artificial Intelligence Civil Rights Act of 2025
The federal S. 3308 (Artificial Intelligence Civil Rights Act of 2025) would bar developers and deployers from offering or using a covered algorithm in a way that causes or contributes to disparate impact or otherwise discriminates based on a protected characteristic in consequential actions such as employment, education, housing, health care, credit, insurance and government benefits. It would require independent-auditor evaluations and annual assessments, public disclosures, human alternatives and appeals, with enforcement by the FTC, State attorneys general and private suits. Introduced December 2, 2025 and referred to the Senate Commerce Committee; no further action.
Algorithmic Accountability Act of 2025
The federal H.R. 5511 (Algorithmic Accountability Act of 2025) would direct the FTC to issue regulations within 2 years requiring covered entities to assess the impact of covered algorithms before and after deployment and file summary reports. Covered entities include those with over $50,000,000 in average annual gross receipts or $250,000,000 in equity value, or data on over 1,000,000 consumers, households or devices. Violations would be treated as FTC rule violations, and state attorneys general could sue. Introduced September 19, 2025 and referred to the House Energy and Commerce Committee; no further action.
Restricts the use by an employer or an employment agency of electronic monitoring or an automated employment decision tool to screen a candidate or employee for an employment decision unless such tool has been the subject of a bias audit within the last year and the results of such audit have been…
New York's A 8328 would have added Labor Law § 203-g restricting employer use of electronic monitoring and automated employment decision tools: bias audits before use and annually, ten business days' notice to employees and candidates, human oversight, a ban on facial, gait or emotion recognition, a private right of action and civil penalties. Introduced December 13, 2023 and referred to the Labor Committee; the enacting clause was stricken January 10, 2024, and it died at the end of the 2023-2024 session.
AN ACT relating to the protection of children using the internet.
Kentucky's SB 345 (2024 session) would have required businesses that knowingly process children's personal information to protect children (under 18) online. Before offering new online services likely to be accessed by children, they would have had to complete data protection impact assessments, set default privacy settings to a high level, and not profile children by default where an assessment found unmitigated high risk. Profiling meant automated processing of personal information to evaluate a person. Violations would have been unfair or deceptive practices with Attorney General penalties of up to $2,500 or $7,500 per affected child. Referred to committee March 1, 2024; died.
Consumer Protection – Online Products and Services – Data of Children (Maryland Kids Code)
Maryland's SB 571 (Kids Code) covers for-profit firms doing business in Maryland that have over $25M revenue, buy, receive, sell or share data of 50,000+ consumers, households or devices yearly, or get at least half their revenue from data sales. For online products reasonably likely to be accessed by under-18s they must do impact assessments (existing ones by April 1, 2026) and set high-privacy defaults. AI-related: no default profiling of children without safeguards and need or a compelling reason; assessments weigh algorithm harms. Fines up to $7,500 per child per violation. Signed May 9, 2024 (Chapter 460; same as HB 603); effective October 1, 2024. Not enjoined; NetChoice suit pending.
Ai, Deepfakes, Cybersecurity, Data Xfers
Alaska's HB 306 would have required a statement ('This communication has been manipulated or generated by artificial intelligence') on campaign communications containing a deepfake meant to injure a candidate or party or deceive voters. It would also have set rules for state agency AI used for consequential decisions: inventories, impact assessments, notice, consent, appeals and bans. It was heard and held in House State Affairs March 19, 2024 and died when the 33rd Legislature ended.
Algorithmic Accountability Act of 2025
The federal S. 2164 (Algorithmic Accountability Act of 2025) would direct the FTC to issue regulations within 2 years requiring covered entities that deploy automated decision systems in critical decisions to assess their impact before and after deployment and file summary reports. Covered entities include those with over $50,000,000 in average annual gross receipts or $250,000,000 in equity value, or data on over 1,000,000 consumers, households or devices. Violations would be treated as FTC rule violations; state attorneys general could sue. Introduced June 25, 2025 and referred to the Senate Commerce Committee; no further action.
Limit government use of AI systems
HB 178 (Chapter 427 of 2025) limits government AI use. State and local government entities and state officers may not use an AI system for cognitive behavioral manipulation, to classify people in ways that cause unlawful discrimination or disparate impact, for a malicious purpose, or to surveil public spaces except to find a missing, endangered or wanted person or as the facial surveillance law allows. AI-produced material published without review by a trained human must be labeled, and public-facing AI interfaces disclosed. Any AI recommendation or decision that could affect a person's rights must be reviewed by a responsible human who can reject or modify it. Effective October 1, 2025.
Creates standards for independent bias auditing of automated employment decision tools.
New Jersey's S2964 would bar an employer or employment agency from using an automated employment decision tool (AEDT) if more than one year has passed since its last independent bias audit. Audits would have to compute selection or scoring rates and impact ratios by sex, race/ethnicity and intersectional categories. Employers would have to post audit results and notify candidates at least 10 business days before use. The text sets no penalties. Introduced March 18, 2024. A bill of the Legislature's 2024-2025 session, which ended January 13, 2026.
Automated Employment Decision Tools (AEDT) Law
New York City's Local Law 144 of 2021 bars employers and employment agencies from using an automated employment decision tool (AEDT) to screen candidates or employees for hiring or promotion in the city unless an independent auditor did a bias audit of it no more than one year before use and a summary of the results and the tool's distribution date are on their website. City residents must get notice at least 10 business days before use. Fines: up to $500 for a first violation, $500 to $1,500 for each subsequent one. Became law without the Mayor's signature in December 2021; effective January 1, 2023; the Department of Consumer and Worker Protection began enforcing it July 5, 2023.
Colorado Artificial Intelligence Act
Colorado's SB 24-205 (the Colorado AI Act) covers high-risk AI that makes, or is a substantial factor in making, consequential decisions such as hiring, lending, housing, insurance and health care. Developers and deployers must use reasonable care against algorithmic discrimination; deployers need risk programs, impact assessments and notices. Signed May 17, 2024 (Chapter 198). Duties run from June 30, 2026 (SB 25B-004) until SB 26-189 replaces Part 17 on January 1, 2027. Under a stipulated April 27, 2026 order in X. AI LLC v. Weiser (D. Colo.), the Attorney General may not start enforcement over violations through 14 days after a ruling on xAI's planned preliminary-injunction motion.
Concerning the use of facial recognition services.
Washington's Chapter 257, Laws of 2020 (ESSB 6280) governs state and local government use of facial recognition services (chapter 43.386 RCW). Before use, an agency must give notice to its legislative authority and publish an accountability report after public comment and community meetings. Decisions with legal or similarly significant effects require meaningful human review; services must be tested in operational conditions and vendors must allow independent accuracy and bias testing. Ongoing surveillance, real-time identification and persistent tracking need a warrant, exigent circumstances or a court order. The governor vetoed section 10 (a task force). Effective July 1, 2021.
Labor and Employment - Automated Employment Decision Tools - Prohibition
Maryland's HB 1255 (2024) would have barred employers from using automated employment decision tools to screen applicants or decide terms of employment unless the tool passed a yearly impact assessment finding no high-risk action, and required employers to notify affected applicants within 30 days of use. Civil penalties would start at up to $500 per violation. An Economic Matters hearing was set for March 6, 2024; it went no further and died when the session ended.
Human Control of AI Act
Illinois's HB 4980 (104th General Assembly) would create the Meaningful Human Control of Artificial Intelligence Act. Public employers and their contractors could not use or procure automated decision-making systems for specified functions without meaningful and continuing human review, and would have to run impact assessments at least every 2 years. It passed House committee as amended; re-referred to Rules April 17, 2026.
AN ACT CONCERNING AUTOMATED DECISION SYSTEMS PROTECTIONS FOR EMPLOYEES.
Connecticut's SB 435 (2026) would have set rules for employers that use automated employment-related decision processes. Deployers would have had to give applicants and employees notices, a statement of reasons for adverse decisions with a chance to correct data and appeal to human review, and an annual independent bias audit; violations of sections 3 to 8 would have been unfair trade practices enforced solely by the Attorney General; aggrieved workers could also have sued. It also covered state-agency AI use. After a Judiciary Committee favorable report, it was tabled for the Senate calendar April 20, 2026 and never voted on; it died when the session adjourned May 6, 2026.
Restricts the use by an employer or an employment agency of electronic monitoring or an automated employment decision tool to screen a candidate or employee for an employment decision unless such tool has been the subject of an impact assessment within the last year; requires notice to employment…
New York's A 9315-B (the "Boundaries on Technology Act") would add Labor Law Article 36 (§§ 1010-1016). An employer with one hundred or more employees could not use an automated employment decision tool unless it had an impact assessment by an impartial auditor within the prior year, repeated annually, with results sent to the Department of Labor for a public registry within 60 days. Employers would have to notify employees and candidates of the tool. Civil penalties would run up to $1,000, $2,000 and $3,000. Amended and recommitted to Ways and Means May 31, 2024; it never passed either chamber and died at the end of the 2023-2024 session.
Establishes criteria for the use of automated employment decision tools; provides for enforcement for violations of such criteria.
New York's S 5641-A (2023-24 session, Sen. Comrie) would have added Labor Law article 20-D on automated employment decision tools (AI that is the controlling factor in a decision to hire, promote or terminate). Deployers and developers would have had to perform annual impact assessments and keep an AI governance program, and deployers to notify affected individuals; the Attorney General could investigate and would give 45 days' notice to cure. Referred to Labor March 10, 2023 and amended and recommitted to Labor January 8, 2024; it died when the session ended.
Fostering artificial intelligence responsibility
Massachusetts H77 would add G.L. Chapter 149B limiting employers' electronic monitoring and use of automated decision systems (independent impact assessments, 10 business days' notice, human oversight, no primary reliance for hiring, discipline or pay), and bar state agencies from using such systems for benefits or rights decisions unless authorized by law. Reported favorably and referred to House Ways and Means December 24, 2025; it has not passed either chamber.
Regulating high-risk artificial intelligence system development, deployment, and use.
Washington's HB 2157 (Substitute House Bill, 2026 session) would have required developers and deployers of high-risk AI systems, those specifically intended to make, or be a substantial factor in making, consequential decisions such as employment, lending, housing, insurance, health care and education, to use reasonable care against algorithmic discrimination, with developer documentation, deployer risk programs and impact assessments, consumer notices, and private lawsuits. It died when the session adjourned March 12, 2026, after the House Rules "X" file on February 19.
Restricts the use by an employer or an employment agency of electronic monitoring or an automated employment decision tool to screen a candidate or employee for an employment decision unless such tool has been the subject of an impact assessment within the last year; requires notice to employment…
New York's S 7623-C (2023-2024) would have added Labor Law Article 36, the "Boundaries on Technology Act" (sections 1010 to 1016). Employers with one hundred or more employees could not use an automated employment decision tool unless an impartial party completed an impact assessment no more than one year before use, repeated annually, with results sent to the Department of Labor for a public registry; employers could not rely solely on tool output. Version C dropped the electronic monitoring rules of earlier versions. Amended May 31, 2024, it never passed either chamber and died when the 2023-2024 session ended.
Enacts the "New York artificial intelligence consumer protection act", in relation to preventing the use of artificial intelligence algorithms to discriminate against protected classes.
New York's A 768, the "New York artificial intelligence consumer protection act," would add General Business Law Article 45-A. Beginning January 1, 2027, developers and deployers of high-risk AI decision systems used in consequential decisions (education, employment, lending, government services, health care, housing, insurance, legal services) would have to use reasonable care against algorithmic discrimination, with documentation, impact assessments, consumer notices and appeal rights. Only the Attorney General would enforce it. Referred to Consumer Affairs and Protection January 7, 2026; it has not passed either chamber.
Health care services: artificial intelligence.
California's SB 503 targets bias in clinical AI. Developers and deployers (health facilities, clinics and medical practices) of clinical decision support systems, AI that aids decisions on timing of care, diagnosis or treatment, must make reasonable efforts to find systems with known or foreseeable risk of biased impacts based on protected characteristics. Developers must mitigate those risks and give deployers a statement of uses and risks plus documentation on training data, bias testing and monitoring; deployers must monitor the systems and take reasonable steps to mitigate bias. Signed September 30, 2026 (Chapter 857); effective January 1, 2027.
Regulates the use of artificial intelligence in aiding decisions on rental housing and loans; requires a study on the impact of artificial intelligence and machine learning on housing discrimination and redlining.
New York's A 3930 would regulate algorithms in housing and lending. Landlords and banks using automated decision tools to screen applicants would need annual disparate impact analyses, 24-hour notice to applicants, and (landlords) public posting of results; landlords could not use pricing algorithms with nonpublic competitor data to set rent. Tenants could sue for damages. The housing agency would study AI and housing discrimination. Would take effect immediately. Referred to Housing January 30, 2025 and January 7, 2026; it has not passed either chamber.
An act relating to restricting electronic monitoring of employees and the use of employment-related automated decision systems
Vermont's H.262 (2025-2026 biennium) would have limited employers' electronic monitoring of employees and use of automated decision systems (ADS) in employment decisions (new 21 V.S.A. sec. 495q). Employers would have had to give 15 days' notice before monitoring and could use an ADS only with human corroboration, an impact assessment and notice, never relying on its output alone. Facial, gait, voice and emotion recognition would be banned. It was referred to House General and Housing on February 19, 2025 and died.
Artificial Intelligence Civil Rights Act of 2025
The AI Civil Rights Act would bar companies from developing or using AI-driven 'covered algorithms' in decisions about jobs, education, housing, utilities, health care, credit, insurance, criminal justice, elections, government benefits or public accommodations in a way that causes a disparate impact or otherwise discriminates based on a protected characteristic. Where harm is plausible, an independent auditor must evaluate the algorithm before deployment, and deployers must assess impacts every year, with summaries made public. The FTC would set rules for human alternatives and appeals. Enforced by the FTC, state attorneys general and private lawsuits. Introduced December 2, 2025.
Artificial intelligence by public bodies; Joint Commission on Technology and Science to analyze use.
Virginia SB 487 (2024) directs the Joint Commission on Technology and Science, with stakeholders, to analyze the use of artificial intelligence by public bodies and whether to create a Commission on Artificial Intelligence. The analysis covers policies for procuring, implementing and continually assessing AI systems, preventing unlawful discrimination and disparate impact, requiring impact assessments, and the possible commission's makeup and duties, with a report to the money and technology committees by December 1, 2024. It incorporates SB 621. Effective July 1, 2024.
Provides disparate impact based on automated decision system as cause of action for certain consumers.
New Jersey's S 4279 (as introduced) would bar a covered entity (lender, housing provider, screening company, creditor, servicer, insurer or their agents) from using an automated decision system in housing or credit decisions in a way that results in a disparate impact on a protected class, a violation of the Law Against Discrimination. It would require impact assessments at least every two years, notice and human review on request, and would let the Division on Civil Rights or Attorney General impose fines of at least $5,000 for certain violations. Introduced May 14, 2026 and referred to the Senate Commerce Committee; not passed.
State-by-State Regulation of AI Bias Audits & Requirements
Artificial intelligence regulation is advancing at the state level. This page lists the 94 published bills we track on AI Bias Audits & Requirements. 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 AI Bias Audits & Requirements 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.