Trump Administration Weighs Liability Rules for Harmful AI

The Trump administration is using existing laws to address AI safety and is considering whether AI developers or the businesses deploying their systems should bear financial responsibility when models cause harm. A new AI task force warned companies to report security incidents and fix harm quickly, while legal experts question whether current cybersecurity and product liability rules can resolve accountability for autonomous AI agents.

EXPANDED:

The administration is using existing statutes rather than new AI rules. Treasury Secretary Scott Bessent and former White House AI adviser David Sacks favor developer financial accountability; Trump said the Justice Department could intervene if systems become uncontrollable. A newly launched Super Intelligence Force warned firms to disclose incidents and fix harm promptly after an Anthropic breach.

Legal uncertainty persists. Experts including CNAS’s Ben Hayum question whether model makers or users bear responsibility when autonomous agents act. The FTC is examining OpenAI, Anthropic and others; Senators Hawley and Murphy and Representative Trahan propose liability changes. Industry views differ: NVIDIA’s Jensen Huang and Sacks lean toward developer responsibility, while OpenAI’s Sam Altman and Anthrop