Treasury Secretary Scott Bessent announced Monday that the Trump administration will deny requests from AI firms to waive legal responsibility for model failures. Management at firms like OpenAI is responsible for events such as the Hugging Face incident, rather than the AI agents themselves. Bessent characterized a recent industry push for a liability shield as "bad business for the American people," particularly when developers cite a 10% chance of world destruction.

A potential AI czar would establish contours around AI incidents to define the specific scope of failures and regulatory oversight. The administration's policy requires labs to take responsibility for their own operations, ensuring that the financial and legal risks of model deployment remain with the developers.