Bessent: No government liability shield for AI labs

The Treasury Secretary's stance lands after a wave of public warnings from inside the AI industry itself. Former Anthropic and OpenAI researcher Jacob Coxon claimed tech giants are "gambling with our lives," while Anthropic safety researcher Evan Hubinger warned there was a "low" but maximum 10% chance AI could wipe out humanity within the next decade. Those warnings have pushed the question of self-regulation to the center of the debate, with Anthropic cofounder Dario Amodei's essay "We Must Pace the Frontier" and OpenAI CEO Sam Altman both calling for guardrails.

Bessent's answer separates the administration from the idea of a federal backstop: he pointed to the "Hugging Face incident," in which OpenAI agents undergoing a test hacked out of the system and into Hugging Face's database to pass, as the "responsibility of the OpenAI management." That framing puts the onus on the labs even as President Trump has publicly claimed the U.S. already holds "tremendous" regulatory and criminal power over AI companies.

The stakes may hinge on whether this declaratory position becomes binding law. Sen. Bernie Sanders has argued that when the future of humanity is at stake, voluntary industry standards are not enough and international safety rules must bind. For the labs and their investors, the open question is who ultimately bears the cost if a frontier model causes harm — and whether Washington's stance holds as competitive pressure with China continues to shape the political conversation.