November 19, 2026


Bipartisan Trio Unveils Federal AI Safety Framework

A bipartisan group of House lawmakers unveiled a framework Thursday for federal oversight of advanced artificial intelligence, proposing transparency requirements, independent evaluations and penalties for developers that violate their own safety commitments.

The Safe and Accountable Federal Evaluations of Artificial Intelligence Act, or SAFE AI Act, is led by Reps. Sara Jacobs (D-Calif.), Celeste Maloy (R-Utah) and Don Beyer (D-Va.). Their proposal arrives as lawmakers debate how to regulate a rapidly developing industry without stifling innovation and as Democrats prepare their AI policy agenda for the next Congress.

“No industry gets to police itself, and AI shouldn’t be the exception,” Jacobs said in a statement, describing the proposal as a “floor — not a ceiling.”

The framework would require AI developers to publish safety policies and could impose civil penalties on companies that violate them. It also envisions third-party evaluators as a tool for government oversight, with strict guidance governing which verifiers may assess developers and their models.

Major technology companies, including OpenAI, have backed similar independent assessment provisions. Some safety advocates, however, question whether evaluators paid directly by AI labs can provide truly independent reviews.

The proposal would also allow judges, at the government’s request, to order measures to mitigate risks from AI systems. Employees at AI labs who speak out about safety concerns would receive whistleblower protections.

Maloy emphasized that the plan “leaves room for Utah and other states to set safeguards of their own.” That approach addresses a major point of contention in the AI debate: Democrats have broadly opposed Republican efforts to preempt state laws, while some technology companies argue that a patchwork of state rules is difficult to navigate. Safety groups generally favor allowing states to adopt stronger protections than the federal government.

The framework joins other bipartisan efforts to define Washington’s role in AI oversight, including the FRONTIER Act from Reps. Lori Trahan (D-Mass.) and Jay Obernolte (R-Calif.). Trahan also released a liability proposal last month amid concerns about who should be held responsible if autonomous AI agents cause harm online or in the physical world.

POLITICO first reported details of the Jacobs-Beyer framework last month.