PRIA Logo™
119-s-5471119th CongressWALLET

AI Systems Transparency Act

Sponsored By: Senator Coons, Christopher A. [D-DE]

Introduced

Summary

Public disclosures and oversight of large AI systems. This bill would create the AI Systems Transparency Act to force developers and deployers of widely used or minor‑facing generative models to publish searchable, plain‑language consumer disclosures and detailed researcher disclosures, and to submit models and deployments to a federal designation and oversight regime.

Personalized for You

How does this bill affect your finances?

Personalize government policy and PRIA will tell you what this bill means for your household, plus every other piece of legislation we track. PRIA reads each provision against your financial profile to show you exactly what matters to your wallet.

Free to download

Bill Overview

Analyzed Economic Effects

5 provisions identified: 1 benefits, 0 costs, 4 mixed.

FTC enforcement and penalties apply

If enacted, violations of the bill's disclosure requirements would be treated as unfair or deceptive acts under the Federal Trade Commission Act. The FTC would be able to enforce the rules using its existing powers, penalties, and remedies. The Commission would have up to one year to write any additional rules needed to carry out the Act. This would increase regulatory enforcement over covered AI systems.

Public disclosures for large AI

If enacted, the Commission would write rules within 150 days requiring covered model developers and deployers to publish two disclosures. One disclosure would be consumer-facing and one would provide detailed technical information for independent researchers. The disclosures would have to be accurate, searchable, plain-language, available via API, and include safety steps, data-use practices, benchmark results, risk-domain evaluations (including CBRN and cyber), and usage summaries like refusal rates and moderation actions. Developers would also have to update disclosures for significant changes and provide dataset dictionaries and testing details.

Rules for model versions and designation

If enacted, the Commission would issue guidance within 150 days on when a model counts as a new version. The Commission would also set rules to decide when a deployment meets the covered user thresholds. The agency could designate a model or system as "covered" if it poses significant risk to minors or the public. Developers and deployers would get at least 30 days written notice to respond, but the Commission could make an immediate designation for imminent risk and must then allow a 30-day response and a 60-day review.

Which AI systems are covered

If enacted, the bill would define which AI systems and operators must follow the rules. A "covered deployer" would be an operator with at least 1,000,000 unique U.S. monthly users in most months. If a service lets minors create accounts, the threshold would be 100,000 unique monthly U.S. users in most months. A "covered model" would include public website, app, API, cloud license, or public release (weights/architecture) models, and a "covered system" would be a model or models plus integrated software that work together.

Technical help for small platforms

If enacted, the Commission would offer technical assistance to covered platforms to help them meet the bill's disclosure and safety rules. The help would focus on small and mid-sized platforms, as the Commission defines them. Only covered platforms that qualify would get this support. The goal is to lower the cost of following the disclosure and safety requirements.

Sponsors & CoSponsors

Sponsor

Coons, Christopher A. [D-DE]

DE • D

Cosponsors

  • Sen. Britt, Katie Boyd [R-AL]

    AL • R

    Sponsored 9/23/2026

  • Sen. Schatz, Brian [D-HI]

    HI • D

    Sponsored 9/23/2026

  • Sen. Lankford, James [R-OK]

    OK • R

    Sponsored 9/23/2026

Roll Call Votes

No roll call votes available for this bill.

View on Congress.gov
Back to Legislation