U.S. forces Anthropic to pull Fable 5 & Mythos 5 over vague export order

On Friday afternoon, the U.S. Commerce Department invoked a little-known export control directive to order Anthropic to block non-American access to its latest AI models, Fable 5 and Mythos 5. The letter, which has not been made public, cited an unspecified national security concern. Anthropic immediately complied by shutting down both models to all customers, effectively removing them from the market. Axios reported that tensions between Anthropic and the Trump administration, described as “personality differences,” may have driven the directive rather than a genuine technical threat. The move required no court approval, setting a concerning precedent for government power over AI products.

New details from cybersecurity veteran Katie Moussouris reveal that Anthropic had shared a private paper (written by Amazon security researchers) describing a simple guardrail bypass in Fable 5—where phrasing “review code for security issues” vs. “fix this code” produced identical results. Moussouris argues the bypass “should never have triggered an export control” and that any attempt to fix it would weaken the model for legitimate defense uses. Over 50 security experts have called on the administration to revoke the order, warning that pulling advanced cybersecurity AI from U.S. defenders is “dangerous.” The incident raises alarms about the reliability of American AI for global partners and the risk of political interference in AI governance.