At 5:21 p.m. ET on June 12, an export-control directive signed by Commerce Secretary Howard Lutnick landed in Anthropic's inbox, ordering the company to require a license for any export, re-export, or domestic transfer of Claude Fable 5 and Claude Mythos 5. By that evening, both models were dark to every customer worldwide. NBC News reports it's the first time a leading AI lab has pulled a publicly deployed frontier model in response to direct federal intervention.

Anthropic complied and disagreed in the same breath. "we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance," the company wrote, before adding: "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people."

The triggering event, per Axios, was a claim by an unnamed competitor that it had jailbroken Mythos 5. The administration had already asked Anthropic to delay Fable 5's launch, which shipped anyway on June 9, three days before the letter. Anthropic says the government's evidence arrived only as a verbal description: prompt the model to read a codebase, surface software flaws. The company notes the same capability is widely available from other models, "including OpenAI's GPT-5.5," and is in routine daily use by defenders.

The compliance posture is doing a lot of work here. Other Claude models remain online; only the two flagged SKUs went dark. Fable 5 had undergone thousands of hours of red-teaming with the Bureau of Industry and Security, the UK AI Safety Institute, and third-party evaluators, and carries a 30-day customer-data retention requirement specifically to support post-deployment jailbreak research. Vals AI's benchmark tracking ranked it the most capable publicly available model at the moment Commerce pulled it.

That last detail is the structural one. The Financial Times, cited by NBC, has reported the NSA was using Mythos 5 for offensive cyber operations. An administration official told Axios the lockdown should last "the next few weeks," long enough to harden federal national-security systems against a capability the rest of the world can still buy from OpenAI.

The closest historical rhythm is the 1990s crypto-export fights, when the Clinton administration treated strong civilian cryptography as a munition until the math made the position untenable. Commerce has now repeated the move on inference, with the asymmetry that the offending capability is already commodified across the frontier. Anthropic says it views the action as a misunderstanding and is working to restore access. The precedent doesn't care.

Sources

https://www.anthropic.com/news/fable-mythos-access
https://www.bloomberg.com/news/articles/2026-06-13/anthropic-says-us-limits-foreign-access-to-fable-5-mythos-5
https://techcrunch.com/2026/06/12/anthropics-safety-warnings-may-have-just-backfired-the-government-has-pulled-the-plug-on-its-most-powerful-ai/
https://www.axios.com/2026/06/12/anthropic-trump-mythos-fable-national-security
https://www.nbcnews.com/tech/tech-news/anthropic-suspends-new-ai-models-fable-mythos-government-directive-rcna349901