Proposed US Senate Bill Would Require AI Firms to Prove Safety Precautions
US senators are negotiating a draft AI regulation bill that would require AI developers to demonstrate a 'duty of care' by proving they took reasonable steps to prevent their models from causing harm. Under the proposal, the US Commerce Secretary could order government audits of AI tools and work to block unsafe model releases. This legislation reflects growing momentum in Washington to establish federal oversight over artificial intelligence safety as advanced models rapidly evolve. If enacted, it could establish a national framework for AI risk governance while preempting fragmented state-level regulations. The draft bill includes provisions allowing companies to challenge government release bans in federal court, while potentially prohibiting individual states from enforcing their own AI risk laws. However, negotiators emphasize that the bill is in early stages with a low immediate likelihood of passing Congress.
## BACKGROUND
'Duty of care' is a legal obligation requiring organizations to adhere to a standard of reasonable care to prevent foreseeable harm to others. As frontier AI models gain advanced capabilities, lawmakers are debating whether government safety audits and mandatory risk controls should be required before major AI tools are released to the public.