PRIA Logo™
S5576119th CongressWALLET

Artificial Intelligence Risk Management and Security Act of 2026

Sponsored By: Senator Warner, Mark R. [D-VA]

Introduced

Summary

Creates a permanent Artificial Intelligence Safety Board at the Department of Commerce to oversee risks from high‑capability “frontier” AI. It pairs mandatory Model Safety Plans and 45‑day pre‑release access with incident reporting and public tracking to spot recurring AI flaws.

Show full summary
  • Developers and AI companies would have to publish a Model Safety Plan for each model and submit capabilities and risk assessments using the Board’s framework. They must give the Board pre-release access to model weights and runtime data at least 45 days before commercial introduction and face civil penalties up to $250,000 per violation counted daily.
  • Operators of critical infrastructure that use AI in industrial control systems would need to report confirmed AI safety or security incidents within 30 days and within 72 hours if the incident poses an imminent threat to national security, infrastructure, or public safety.
  • Researchers, independent testers, and private firms would gain access to Secure Research Test‑Beds hosted by agencies such as the National Security Agency and Department of Energy National Laboratories, with cost‑recovery access and support from the National Institute of Standards and Technology (NIST).
  • NIST and the Cybersecurity and Infrastructure Security Agency (CISA) would set up voluntary sharing mechanisms within 180 days and a publicly accessible incident database within 1 year that anonymizes parties unless they consent to naming.
  • The Board would propose technical standards within 90 days of formation and receiving agencies would adopt or explain modifications within 30 days. NIST would also develop an Agentic AI Profile and a common template for documenting AI agents within 18 months.

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

3 provisions identified: 1 benefits, 1 costs, 1 mixed.

New AI safety rules for developers

If enacted, the bill would require the Secretary of Commerce to create an Artificial Intelligence Safety Board within 30 days. The Board would propose technical standards within 90 days, and agencies would have 30 days to adopt or modify them. Developers would have to follow applicable standards and file Model Safety Plans when they create, substantially modify, or use models for training or evaluation. Frontier model developers would also have to give the Board pre‑release access at least 45 calendar days before introducing a model into interstate or foreign commerce. Violations could lead to civil penalties up to $250,000 per violation, with each day of a continuing violation treated as a separate violation, and the Attorney General could seek injunctions.

AI incident reporting and database

If enacted, the bill would require NIST, working with CISA, to set up a voluntary, confidential way to share AI security incidents, safety incidents, and identified AI flaws within 180 days. NIST would also create a public database and flaw catalog within one year and review submissions for inclusion. Information would be anonymized unless consent is given, protected from certain FOIA disclosures, and developers would be consulted before publication. Developers of frontier models and some critical infrastructure operators would also have to file confirmed incident reports within 30 days, or within 72 hours if the incident poses an imminent threat to national security, critical infrastructure, or public safety.

Tech guidance and secure AI testbeds

If enacted, the bill would require NIST to develop an Agentic AI Profile and begin a common AI‑agent documentation template within 18 months. NIST would also provide nonbinding risk guidance within 180 days. The Secretary could arrange secure computing environments from federal partners and make them available to private researchers, agencies, and qualified experts on a cost‑recovery basis for secure testing and evaluation.

Sponsors & CoSponsors

Sponsor

Warner, Mark R. [D-VA]

VA • D

Cosponsors

  • Sen. Schatz, Brian [D-HI]

    HI • D

    Sponsored 9/29/2026

  • Sen. Kim, Andy [D-NJ]

    NJ • D

    Sponsored 9/29/2026

Roll Call Votes

No roll call votes available for this bill.

View on Congress.gov
Back to Legislation