Anthropic is briefing the U.S. government on Claude Mythos despite its ongoing supply-chain risk dispute with the Pentagon

Anthropic co-founder Jack Clark said at a Semafor conference that the company is in conversations with the U.S. government about Claude Mythos, its unreleased AI model the company has warned is too dangerous to release publicly, saying Anthropic's position is that “the government has to know about this stuff.” Clark characterized the ongoing Pentagon dispute as a “narrow contracting dispute” that he says won't conflict with Anthropic's national security concerns, though he did not elaborate on whether the conversations include any discussion of removing the supply-chain risk designation. The Pentagon designated Anthropic a supply-chain risk in February after Defense Secretary Pete Hegseth demanded unrestricted access to its models, a designation a federal appeals court declined to lift last week.