Anthropic’s safety warnings may have just backfired — the government has pulled the plug on its most powerful AI
TL;DR
The U.S. government ordered Anthropic on June 12 to immediately disable Claude Fable 5 and Claude Mythos 5. Other Claude models remain available. The order is framed as export control for foreign nationals. In practice, Anthropic took both models offline worldwide because it says narrower blocking was not feasible. Mythos was Anthropic's strongest cybersecurity model and had been shared only with about 50 vetted partners, including Amazon, Apple, Google, Microsoft, and CrowdStrike.
Nauti's Take
Anthropic has reason to push back if the government's evidence is really only verbal and narrowly scoped. But the company also spent years telling the market that its models carried unusual risks and that its safety culture was the big differentiator.
It looks weak when the same company suddenly argues that a concrete intervention is just about an ordinary jailbreak. Safety is a useful brand story until regulators treat it as an operational claim.
Briefingshow
Anthropic built its brand around safety and described Mythos as too powerful for broad release. That same warning language now gives regulators a reason to intervene aggressively. For teams, frontier AI will not be judged only by benchmarks, but by political risk, exportability, and whether access can be controlled precisely.
Tweets
Sources