Policy

House trio proposes SAFE AI Act with federal safety rules

Bipartisan framework would mandate transparency, independent evaluations, and civil penalties while preserving state regulatory authority.

Omega Editorial · November 19, 2026 · 3 min read

Key takeaways

  • The SAFE AI Act would create federal safety requirements for frontier AI systems, including mandatory transparency disclosures and civil penalties for violations.
  • The bipartisan framework explicitly preserves state authority to impose stricter AI regulations, addressing a key partisan divide in previous legislative efforts.
  • Independent third-party evaluators would assess AI models under strict government guidance to ensure genuine independence from the companies they evaluate.
  • The legislation includes whistleblower protections and grants judges authority to order mitigation measures for risky AI systems.
  • The proposal joins multiple bipartisan AI bills introduced this Congress as lawmakers race to establish federal oversight of the technology.

House lawmakers introduce federal AI safety framework

Three members of Congress have introduced legislation that would establish the first comprehensive federal safety requirements for frontier artificial intelligence systems. The Safe and Accountable Federal Evaluations of Artificial Intelligence Act—or SAFE AI Act—represents a bipartisan effort to create baseline standards for the rapidly evolving industry.

Reps. Sara Jacobs (D-Calif.), Celeste Maloy (R-Utah), and Don Beyer (D-Va.) unveiled the framework Thursday, joining a growing list of legislative attempts to define Washington's role in AI oversight.

Why it matters

The proposal addresses a fundamental tension in AI policy: how to establish consistent federal standards without blocking states from imposing stricter protections. By explicitly preserving state authority, the framework attempts to bridge a partisan divide that has stalled previous regulatory efforts. The approach could provide a template for future AI legislation as both parties prepare for policy debates in the next Congress.

Key provisions of the framework

The legislation would require AI developers to publish detailed safety frameworks outlining how their systems are designed and tested. Companies that violate these published standards would face civil penalties, creating enforceable consequences for safety failures.

A central feature involves independent third-party evaluators who would assess AI models and their potential risks. However, the bill includes strict guidance on which evaluators can perform these assessments—a response to concerns that evaluators paid directly by AI companies may lack true independence.

The framework would also grant judges authority to order mitigation measures for AI systems when requested by government agencies. Additionally, it establishes whistleblower protections for employees at AI laboratories who raise safety concerns.

Balancing federal and state authority

Unlike some Republican proposals that would preempt state AI regulations, the SAFE AI Act explicitly allows states to set their own safeguards. Rep. Maloy emphasized this feature, noting the framework "leaves room for Utah and other states to set safeguards of their own."

This approach reflects Democratic opposition to federal preemption, while some technology companies have argued that complying with multiple state regulatory regimes creates operational challenges. Safety advocacy groups generally support allowing states to exceed federal baseline requirements.

Broader legislative landscape

The SAFE AI Act joins other bipartisan efforts this Congress, including the FRONTIER Act introduced by Reps. Lori Trahan (D-Mass.) and Jay Obernolte (R-Calif.). Rep. Trahan also released a separate liability proposal last month addressing accountability questions if autonomous AI agents cause harm.

Rep. Jacobs framed the legislation as establishing minimum standards rather than comprehensive regulation. "No industry gets to police itself, and AI shouldn't be the exception," she said, describing the proposal as "a floor—not a ceiling."

The framework details were first reported by POLITICO last month, with the formal introduction occurring as Democrats prepare their AI policy agenda for the upcoming Congress.

  • #ai regulation
  • #safe ai act
  • #congress
  • #ai safety
  • #bipartisan legislation
  • #federal policy

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in Policy