Fable 5 and Mythos 5 have now been dark for eight straight days, taken offline at 5:21 p.m. ET on Friday, June 12 by a Commerce Department export-control directive that an unnamed expert told The Globe and Mail is the first ever issued against an AI model. Asked about it at the G7 in Évian-les-Bains this morning, President Donald Trump told reporters negotiations are “going fine.” That’s the closest thing to an off-ramp the company has gotten in public.
In private, the off-ramp looks more developed. According to The Globe and Mail, Commerce Secretary Howard Lutnick has been holding regular calls with Anthropic officials, and senior Anthropic technical staff met with Commerce in Washington on Monday. A Trump administration official told the paper that Anthropic engineers have been in contact with the government virtually every day since the directive landed.
The stated rationale keeps shrinking. Anthropic’s own statement characterizes the underlying trigger as “verbal evidence of a potential narrow, non-universal jailbreak” tied to “previously known, minor vulnerabilities.” The Wall Street Journal reports the paper’s authors are security researchers at Amazon, and that CEO Andy Jassy raised the findings directly with Treasury Secretary Scott Bessent. Katie Moussouris, the Luta Security founder who was shown a private copy of the paper, told TechCrunch the issues “cannot meaningfully be fixed, and any attempt would only weaken the model for defense.”
The order itself uses authorities from the 2018 Export Control Reform Act and, per a Lutnick letter seen by Reuters, threatens “prompt criminal and civil penalties.” It covers “any foreign national, whether inside or outside the United States, including foreign national Anthropic employees,” which is how a domestic compliance order becomes a global blackout for a company valued at roughly $1 trillion.
The timing hasn’t escaped notice. CNBC points out the directive arrived hours after SpaceX closed its first day of public trading, and days after Dario Amodei published an essay arguing frontier models “should be required to go through technical testing and auditing, and their release should be blocked or reversed.” Daniel Remler, a senior fellow at CNAS, told CNBC that voluntary framework “sure looks mandatory if there are going to be consequences for not doing what the government says.” Anthropic now calls the action a “misunderstanding.”
On Sunday, more than 80 cybersecurity executives, including leaders at Nvidia and Adobe, published an open letter to Lutnick and National Cyber Director Sean Cairncross asking for the order to be lifted. Amodei spent the spring lobbying for exactly the kind of state authority now being used against him. The lesson, drawn unusually fast, is that the auditing regime a frontier lab builds for its competitors tends to arrive at its own door first.
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.theglobeandmail.com/business/article-anthropic-trump-officials-deal-restore-fable-5-mythos-5/
- https://techcrunch.com/2026/06/15/the-us-governments-anthropic-models-ban-was-never-about-an-ai-jailbreak/
- https://www.cnbc.com/2026/06/17/anthropic-ai-regulation-trump.html