At 5:21 p.m. ET on Friday, June 12, Anthropic received a letter from the Commerce Department's Bureau of Industry and Security ordering it to suspend access to Fable 5 and Mythos 5 for any foreign national. Because Anthropic can't filter a user base of hundreds of millions by nationality in real time, it did the only thing the letter operationally permitted: it shut both models off worldwide, three days after launch.

The letter, signed by Commerce Secretary Howard Lutnick and addressed to CEO Dario Amodei, cited national-security authorities but, per Anthropic's public statement, contained no specific technical findings. Bloomberg confirmed the document with a U.S. official. An administration official told Axios the government had earlier tried and failed to delay Fable 5's release, and that the block could last several weeks pending strengthened safeguards.

This is the first time U.S. export-control law has been pointed at a deployed commercial model rather than at chips. The frame is the same one Washington has used on hardware since the October 2022 BIS rules on advanced semiconductors. The object is new.

The factual basis is unusually thin. Anthropic's reconstruction is that the cited concern is a single non-universal jailbreak: ask the model to read a specific codebase and fix software flaws. Lawfare reports the demonstration was performed for officials by Amazon, the "highly credible trusted partner" described by former White House AI czar David Sacks. Anthropic says no tester has produced a universal jailbreak against Fable 5's safeguards across thousands of hours of pre-launch red-teaming with the U.S. government, the UK AI Safety Institute, and third-party reviewers, and points to its 30-day customer-data retention policy as part of a defense-in-depth posture.

In a written statement, Anthropic argued: "We have validated that the level of capability displayed there is widely available from other models (including OpenAI's GPT-5.5), and is used every day by the defenders who keep systems safe."

The company is already in court with the administration. On March 9, after the Department of Defense designated it a supply-chain risk, Anthropic sued in the Northern District of California and the D.C. Circuit. California granted a preliminary injunction against the Claude use ban. The D.C. Circuit declined to stay the designation. The June order arrives into that posture, not into a clean slate.

For enterprise buyers, the operational lesson is structural rather than political. A frontier model can now become unreachable by directive rather than outage, on a Friday evening, with no notice. Multi-model abstraction layers like LemonLime stop being a procurement nicety and start being continuity infrastructure. The kill switch exists. It has been used.

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://www.cnbc.com/2026/06/12/anthropic-disables-access-to-fable-5-and-mythos-5-to-comply-with-government-directive.html
https://www.lawfaremedia.org/article/a-kill-switch-for-frontier-ai
https://www.nextgov.com/artificial-intelligence/2026/06/anthropic-suspends-top-ai-models-after-us-export-control-order/414173/