Anthropic suspended access to Fable 5. Globally. Not just for the export-controlled cases the directive technically targeted. Everyone.
Commerce Secretary Howard Lutnick sent the letter to Dario Amodei at 5:21pm ET on June 12. Export controls on Fable 5 and Mythos 5, immediate compliance required. The “jailbreak” that justified it? Anthropic says the government gave them verbal evidence of a technique that essentially consists of asking the model to read a specific codebase and fix any software flaws. That’s the national security concern. GPT-5.5 does the same thing.
Anthropic reviewed the evidence and said the quiet part out loud: “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
That’s not a defense of Fable 5. That’s a warning about what just became possible.
The irony no one needed
This week Anthropic published a policy paper calling for governments to have authority to block or deter dangerous deployments of frontier AI. Dario published his own essay arguing the evidence now demands “more serious and binding regulation.”
They asked for this mechanism. Then it arrived with no specifics, no published criteria, no third-party testing. Just a verbal briefing and a deadline.
Be careful what you wish for.
The part that matters most
They may bring Fable 5 back quickly, at least for “approved” users, whatever that ends up meaning. But the demonstration has already happened. The US government flipped the switch on a world-leading AI model overnight. “Foreign nationals” won’t have access. That sentence should be sitting uncomfortably with every government, CTO, and AI strategy team outside the United States right now.
The dependency isn’t theoretical anymore. It showed its teeth.
If this isn’t the final wake-up call for Europe and others to make AI a genuine top priority, I don’t know what ever will be.