White House · 2026-05-04 · major
White House Weighs Executive Order to Vet Frontier AI Models Before Public Release — Mythos Cited as Catalyst
Trump administration briefed Anthropic, Google and OpenAI last week on a draft order that would create an AI working group and a formal pre-release government review for frontier models — a sharp reversal of its hands-off stance, triggered by Anthropic's Claude Mythos cyber capabilities.
U.S. government may require frontier labs to submit new AI models for safety review before public release.
What is it?
A draft Trump-administration executive order, first reported by The New York Times and confirmed by Bloomberg and Axios on May 4, 2026, that would establish an AI working group of tech executives and federal officials. The group would design a formal government review process new frontier AI models must pass before release.
How does it work?
White House staff briefed representatives from Anthropic, Google and OpenAI on the proposal last week. One option under discussion gives the government early access to test new models without the power to block release; another is closer to the UK model where multiple agencies certify safety. Final scope will be set by Trump in any actual order.
Why does it matter?
It is a reversal of the administration's earlier deregulation push and the first concrete sign Washington wants pre-release control over frontier AI. The catalyst was Anthropic's Claude Mythos, which the company itself declined to release publicly because of its zero-day discovery capability. If signed, the order would let regulators see — and potentially gate — the next Mythos-class system.
Who is it for?
AI labs, policy teams, security researchers