Senate Negotiators Propose Federal 'Duty of Care' for Advanced AI Models
U.S. Senate aides are debating legislation that would impose strict safety requirements on major tech companies developing high-capability artificial intelligence.
U.S. Senate negotiators are considering legislation to hold advanced AI developers responsible for preventing catastrophic risks, according to two Senate aides involved in the discussions. The proposal establishes a duty of care requiring companies to design products specifically to avoid severe dangers such as the creation of nuclear or biological weapons by bad actors. Federal regulators would gain authority to block the release of unsafe AI models, with affected companies permitted to challenge these decisions in federal court. One aide noted that states attempting to enforce their own regulations regarding these specific risks would be preempted. The measures aim specifically at U.S. firms developing the most advanced models, which currently include Alphabet's Google and Anthropic. The precise structure of the law remains under negotiation.
