1. Bottom Line: No Single Federal “AI Transparency Act” Is Enacted
Multiple bills using the phrase “AI Transparency Act” or similar titles have been introduced in Congress, but none have been enacted into law as of April 2026. The federal legislative landscape on AI transparency is fragmented: different committees have different jurisdiction, different bills have different scope, and the absence of a comprehensive federal framework has accelerated state-level action.
What exists instead is a patchwork of:
- Federal proposals that have not passed (covered in Section 2)
- State laws that are in effect now, covering different aspects of AI transparency (Sections 3–4)
For compliance purposes, the relevant law is almost certainly state law, not federal. Source: Congress.gov (search for “AI transparency” to review introduced bills).
2. Federal Proposals: What Has Been Introduced
Several federal proposals have used AI transparency terminology:
Generative AI Copyright Disclosure Act
Introduced in the Senate and House (multiple versions), this bill would require developers of generative AI systems to disclose copyrighted works used in training data. As of April 2026, it has not passed either chamber. The bill is narrowly scoped to training data and copyright, not broader AI transparency. Source: Congress.gov S.2037.
DEFIANCE Act (2024)
The DEFIANCE Act, enacted in 2024, addresses non-consensual deepfake intimate images — not general AI transparency. It creates a federal civil right of action for victims of AI-generated intimate imagery. Source: Congress.gov S.4569.
AI Transparency in News Act / TRAIN Act
Several bills in the 118th and 119th Congress address AI and media, requiring disclosure when AI-generated content appears in certain contexts. None have been enacted as of April 2026.
3. California CAITA (AB 853): The Most Comprehensive Enacted AI Transparency Law
The closest thing to a comprehensive “AI Transparency Act” is California’s California AI Transparency Act (CAITA), enacted as AB 853 and effective August 2, 2026. CAITA covers:
What CAITA Requires
- Latent disclosure in AI-generated content. GenAI providers must embed machine-readable latent disclosures in AI-generated images, audio, and video. This allows detection tools to identify AI-generated content.
- Manifest disclosure option. Providers must also offer users the option to add visible, human-readable AI disclosure to content.
- AI detection tool. Providers must make an AI-detection tool available to allow users to verify whether content was AI-generated.
- 2027: Platform distribution limits. Starting 2027, platforms may not distribute GenAI systems that lack proper disclosure capabilities.
- 2028: Capture device requirements. Manufacturers of cameras, phones, and other capture devices must embed latent disclosure capabilities by default.
Who It Covers
CAITA covers “providers” of GenAI systems — meaning the developers and operators of AI tools used to create or modify images, audio, or video. It does not cover AI systems that generate text only. Penalties up to $5,000/violation.
4. Other Enacted State AI Transparency Laws: CO, TX, UT, NY
| State | Law | Effective | What It Requires |
|---|---|---|---|
| California | AB 853 (CAITA) | 2026-08-02 | Latent disclosure in GenAI images/audio/video; manifest disclosure option; AI detection tool |
| Colorado | SB 24-205 | 2026-06-30 | Pre-decision consumer notices for high-risk AI; all consumer-facing AI must disclose AI interaction |
| Texas | HB 149 (TRAIGA) | 2026-01-01 | Statewide AI governance with disclosure obligations; AI Council established |
| Utah | SB 149 | 2024-05-01 | GenAI systems must disclose AI use if asked; regulated occupations must proactively disclose |
| New York | S.8420-A | 2026-06-09 | Ads using AI-generated synthetic performers must clearly disclose; $1,000–$5,000/violation |
Colorado SB 24-205
Colorado’s SB 24-205, amended by SB25B-004, is the broadest enacted state AI transparency framework outside California. It requires: pre-decision consumer notices when AI substantially influences a consequential decision (employment, credit, housing, healthcare); disclosure that the consumer interacted with an AI system for all AI systems deployed to consumers; and risk assessments for high-risk AI systems. Penalties up to $20,000/violation.
Utah SB 149
Utah’s SB 149 (2024) was the first enacted AI disclosure law in the country. It requires anyone using GenAI to interact with a person to disclose AI use if asked. Regulated occupations (healthcare, legal, accounting, real estate) must disclose proactively, without being asked. Civil penalty up to $2,500/violation.
Texas TRAIGA (HB 149)
Texas HB 149 (TRAIGA), effective January 1, 2026, establishes a comprehensive AI governance framework for Texas, including disclosure obligations for state agency AI deployments and consumer-facing AI uses. It establishes a state AI Council and includes provisions on biometric identifiers and discrimination.
New York S.8420-A
New York S.8420-A, effective June 9, 2026, requires conspicuous disclosure in advertisements using AI-generated “synthetic performers.” It amends New York’s General Business Law and targets deceptive AI use in advertising — not general AI transparency.
5. Federal vs State: Side-by-Side
| Dimension | Federal | State (CA, CO, TX, UT, NY) |
|---|---|---|
| Enacted law? | No comprehensive AI transparency act | Yes — multiple enacted laws |
| Scope | Narrow proposals (deepfakes, copyright) | Broad: GenAI content, chatbots, high-risk AI |
| Enforcement | N/A (not enacted) | AG enforcement; civil penalties |
| Compliance date | TBD | Already in effect (UT 2024, TX 2026, CA 2026) |
6. What Laws Apply to Your AI System?
The answer depends on (a) what your AI system does and (b) where your users are located:
- GenAI image/audio/video creation tool: California CAITA (AB 853) applies if you serve California users. Effective August 2, 2026.
- Chatbot or conversational AI: Utah SB 149 (2024), Colorado SB 24-205, Texas TRAIGA. Disclosure of AI status is required in all three states.
- High-risk AI (employment, credit, housing, health): Colorado SB 24-205 requires pre-decision notices. Texas TRAIGA applies to certain uses.
- AI in advertising: New York S.8420-A applies to synthetic performer ads. Colorado and California have additional requirements.