Anthropic, the company behind the Claude family of AI models, abruptly took its most advanced "Mythos-class" models offline after what The Verge describes as a "Friday evening ultimatum from the Trump administration."
According to The Verge, the company moved quickly once the demand landed, dispatching a wave of executives to Washington, DC. But two weeks on, the situation has only grown murkier. The Verge reports that public updates have been "suspiciously lacking," with no resolution in sight to the standoff.
The exact nature of the dispute remains unsettled. WION frames the situation as a broader "feud" between Anthropic and the Trump administration, signaling that the conflict extends beyond a single technical decision.
There are signs of a partial thaw. Geeky Gadgets reports that one model, Claude Fable 5, has returned to service following what it calls a "government security scare" — suggesting at least some of the offline models are coming back online, even as the larger negotiation continues.
What the sources make clear is the unusual dynamic: a leading AI company pulling flagship products from the market under direct pressure from the federal government, then negotiating quietly rather than publicly.
Many details are still unconfirmed — including the precise security concerns at issue, the full scope of what was taken down, and the terms of any deal being discussed.
Why it matters: When a government can prompt one of the world's top AI labs to yank its most powerful systems offline overnight, it signals that the relationship between Washington and the AI industry is becoming a direct lever over which tools the public can and cannot use.