What happened
On 19 November 2026, Reps. Don Beyer (D-VA, co-chair of the Congressional AI Caucus), Sara Jacobs (D-CA) and Celeste Maloy (R-UT) released an outline of the forthcoming 'Safe and Accountable Federal Evaluations of Artificial Intelligence (SAFE AI) Act,' a bipartisan bill that would create federal safety requirements for frontier AI. It would mandate safety frameworks, incident reporting and increased transparency from frontier labs with strong government oversight and due process, include civil penalties for violations, and provide targeted emergency authority to stop dangerous models posing a serious and immediate danger. The sponsors stated the proposal is 'a floor, not a ceiling,' expressly empowering states to adopt stronger protections.
Why it matters
This is the most concrete bipartisan frontier-AI safety framework being drafted for the new Congress. It combines mandatory safety frameworks, incident reporting, transparency, civil penalties and government emergency authority over advanced models, and — unlike some competing bills — avoids broad federal preemption of state AI laws. It would materially change compliance obligations for frontier labs if introduced and advanced.
Action needed
Frontier AI developers and safety teams should review the outline and expect formal introduction; prepare for mandatory safety-framework, incident-reporting and transparency obligations plus potential emergency-order exposure.