Anthropic said on June 12, 2026 that its AI models Fable 5 and Mythos 5 could be back within days. The company had shut them down worldwide after a U.S. Commerce Department order. Chris Ciauri, who oversees Anthropic’s international business, called the freeze temporary. Anthropic said the order arrived just three days after launch and required the company to block foreign nationals on national security grounds.
Anthropic said it took both models offline everywhere because checking users by nationality at that scale was hard. Mythos 5 is a restricted cybersecurity model available only to vetted partners. Fable 5 is the public version, with extra safety guardrails built in.
Reports say the order followed Amazon’s discovery of a jailbreak for Fable 5, and came amid concerns that SK Telecom had accessed Mythos 5 through Project Glasswing. U.S. officials were also reportedly worried about possible ties to China.
Anthropic, though, says that reading of the situation came from a misunderstanding. It also says the jailbreak was limited and, broadly speaking, looked a lot like issues that have shown up in other models. If you follow AI policy, this case is worth watching. It may be the first known U.S. export-control action aimed at an AI model itself, not the chips behind it.
That leaves a pile of compliance questions around nationality checks, partner access, and cloud delivery, especially as G7 talks continue and digital-sovereignty debates in Europe, India, and South Korea get more heated. Reports also say about 100 cybersecurity professionals have already called the ban an overreaction.
Once Anthropic restores access, you’ll be able to use Fable 5 publicly through Anthropic, while Mythos 5 will stay limited to vetted partners.
