Two dozen firms sign White House Accord on Super Intelligence

The accord follows a string of AI security incidents that pushed OpenAI to halt training and pause releases. Some signers—Anthropic, Google and OpenAI—had already committed to external audits, and many firms had already volunteered for secretive government safety testing, so the Tuesday deal largely formalizes existing practice. Trump described it as "almost like a constitution," while critics including Sen. Bernie Sanders called it a government "of, by and for the billionaire class."

The agreement leaves key questions open. No details have been released on how the audits would work, and it is unclear who will conduct them; Bloomberg reported the selection process looks politicized and that some candidate reviewers have already met the Center for AI Standards and Innovation. Industry sources told Axios they doubt self-policing will solve the safety problem.

Meanwhile, state authorities and the FTC may apply more concrete pressure. OpenAI made commitments to California and Delaware attorneys general during its restructuring, and both states are investigating whether governance failures contributed to recent loss-of-control incidents. That suggests the real test of AI oversight may sit with state regulators rather than the White House accord.