Full Breakdown
Trump Administration Export Control Order Suspends Anthropic's Mythos and Fable AI Models
6/26/2026, 8:35:40 PM
Export Control Order Halts Anthropic's Mythos and Fable Models
On June 12, the Trump administration issued an export-control order forcing Anthropic to block access to its Mythos 5 and Fable 5 models for any foreign national. The directive shut both models down, a condition that has persisted for two weeks with no public resolution.
Official Statements & Responses
The order cites national-security concerns over the models’ dual-use capabilities. The Department of Commerce had previously tested Fable 5 without objection, but acted after Amazon CEO Andy Jassy flagged a guard-rail bypass. Anthropic has declined comment, saying there is “no news to share.”
Timeline of the Standoff
Stakeholders and Financial Stakes
Anthropic’s current negotiating team includes co-founder Tom Brown and policy chief Sarah Heck, while President Donald Trump’s administration drives the order. Shareholders Google and Amazon watch closely. Mythos-class tokens sell for twice Opus 4.8 rates, funding a $15 billion-per-year compute deal with SpaceX and an imminent IPO. OpenAI’s GPT-5.5 Cyber recently outperformed Mythos 5, prompting the administration to seek a staged release of GPT-5.6 under customer-by-customer approval.
Criticism & Opposition
Katie Moussouris says the reported guard-rail bypass is “significantly overblown” and that allowing AI to fix code is vital for defensive security. Alex Stamos called the order “incredibly stupid,” warning it hampers a U.S. champion as China advances. Cybersecurity leaders argue that, if regulation is needed, this approach is misguided.
Conflicting Reports & Gaps
Sources note the Department of Commerce tested Fable 5 without complaint, yet the export order followed a reported guard-rail bypass. No established framework exists for AI export controls, leaving Anthropic and regulators without a shared checklist. The precise reasons for the stalemate remain undisclosed.
Verbatim Quotes
- “Defenders need to be able to ask AI to fix the bugs in a file, explain why the fix matters, and write tests that confirm the patch works,” — Katie Moussouris, Luta Security
- “That is not a guardrail bypass. It is the most valuable thing an AI model can do for defensive security: executing the find, fix, and test loop defenders run every day.” — Katie Moussouris, Luta Security
- “One of America’s champions is being kneecapped by the US government while we’re in a race with the Chinese. It’s just incredibly stupid.” — Alex Stamos, Corridor
- “any foreign national” — Trump Administration, Export Control Order
What’s Next
Anthropic’s negotiations will determine whether the order expands to other U.S. AI firms. The timing of the company’s IPO and OpenAI’s pending releases hinge on regulatory clearance, while global competitors, especially China, continue to close the AI gap.
