Sen. Cantwell proposes federal AI safety standards and independent audits before advanced models launch
Sen. Maria Cantwell, the top Democrat on the Senate Commerce Committee, released a six-principle framework Wednesday that would have federal experts set enforceable safety standards for the most powerful AI systems, and bar covered models from release until an independent audit confirms they comply. It is a framework, not a bill, and Congress is out until after the election.
What it proposes: Under the framework, the National Institute of Standards and Technology (NIST), working with other federal agencies, would write clear, enforceable, risk-based standards for frontier AI systems that could cause catastrophic harm, according to Cantwell’s office. Those risks include AI-enabled cyberattacks; chemical, biological, radiological and nuclear threats; and loss of human control. Covered models “should not be released until they have undergone an independent audit” confirming they meet the standards.
Testing and disclosure: Government and independent experts would continuously stress-test covered models, and audit firms would examine AI developers much as outside auditors check a company’s financial statements. Developers would have to give testers access, publish plain-language risk disclosures modeled on those public companies give investors, promptly report serious safety incidents, and protect whistleblowers from retaliation.
The legal angle: The framework says developers should remain liable under existing civil and criminal law for foreseeable harms from their systems. It also says people should be able to learn when AI materially affected a consequential decision about them, such as employment, health care or credit, challenge the result, and get review by a qualified human. Other planks cover child safety, worker training and a U.S.–China crisis channel for major AI incidents.
What’s unresolved: Penalties for noncompliance and who the third-party auditors would be are still undecided, The Seattle Times reported. Cantwell told the paper the principles are a starting point for legislation she says is coming. Senate Minority Leader Chuck Schumer and Microsoft President Brad Smith issued supportive statements, the Times said.
Where it stands: Congress has not advanced comprehensive AI legislation this year, and the House and Senate won’t return until after the Nov. 3 midterms, leaving few days for action in 2026, the Times reported. Roll Call and POLITICO also covered the release.
Sources
- U.S. Senate Commerce Committee (Democrats) — Cantwell Outlines Comprehensive Governance Framework for Safe and Secure Frontier AI
- The Seattle Times — Cantwell unveils AI safety framework as Congress stalls ahead of election
- Roll Call — Sen. Maria Cantwell looks to federal testing role in AI safety framework
- POLITICO — Top Commerce Committee Democrat releases AI framework