An unprecedented national security intervention has sent shockwaves through the artificial intelligence industry. Anthropic has abruptly suspended global access to its newly crowned frontier models, Claude Fable 5 and Claude Mythos 5, just days after their public release.
The immediate worldwide freeze follows a sweeping security export control directive issued by U.S. Commerce Secretary Howard Lutnick. The catalyst? A reported security tip-off from researchers at Amazon, who demonstrated that the highly advanced models could be manipulated to bypass native safety protocols.
Why Did Anthropic Shut Down Claude Fable 5?
The sudden global shutdown traces back to research conducted by engineers at Amazon, a primary cloud partner and major investor in Anthropic.
According to reports, Amazon researchers successfully utilized a sequence of complex prompts to “jailbreak” Fable 5. By bypassing its safety guardrails, the researchers forced the model to evaluate code bases and reveal dangerous software vulnerabilities.
Concerned by how easily these cyber-capabilities could be exposed, the findings were escalated directly to the federal government. The primary fear surrounds Mythos 5, an unmitigated model designed to rapidly weaponize flaws—shrinking standard vulnerability exploit timelines from weeks to mere hours.
The Commerce Department Directive: A Global AI Blackout
The Commerce Department’s Bureau of Industry and Security reacted swiftly. On Friday evening, Anthropic CEO Dario Amodei received a formal letter from Commerce Secretary Howard Lutnick enforcing severe export controls.
The directive explicitly bars access to Fable 5 and Mythos 5 for:
Any foreign governments, entities, or companies operating outside the United States.
All foreign nationals currently residing inside the United States, including Anthropic’s own international workforce.
Because Anthropic cannot verify the citizenship or legal nationality of millions of active users in real-time, the company was forced to shut down global access entirely to guarantee full legal compliance with the federal order.
Anthropic Responds: “A Complete Misunderstanding”
Anthropic has complied with the legal mandate under intense protest, issuing an extensive response defending its frontier models. The company claims the government’s aggressive move is a sweeping overreaction based on a narrow, non-universal jailbreak.
| Government Security Stance | Anthropic’s Technical Defense |
| The models possess dangerous, unaligned cyber-capabilities that pose a dual-use threat. | The identified exploits were minor, simple, and already widely known in the cybersecurity industry. |
| Bypassing the guardrails grants malicious actors an immediate strategic advantage. | Other public models (including competitors) surface identical results without needing any bypasses. |
In an official statement, Anthropic noted:
“Perfect jailbreak resistance is not possible for any model provider. We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, it would essentially halt all new model deployments.”
A History of Friction: The White House vs. Frontier AI
This dramatic shutdown is not an isolated incident; it marks the climax of an increasingly turbulent relationship between Anthropic and Washington.
Prior to this export control order, the Trump administration had already repeatedly pressured Anthropic executive leadership to pause or delay the public rollout of its newest tier of models. Relations deteriorated further after the Department of Defense previously flagged the AI firm as a potential supply chain risk when early-stage regulatory negotiations broke down.
While Anthropic’s lower-tier models remain fully operational, the future of the Fable 5 and Mythos 5 infrastructure hangs in legal limbo. The tech giant is actively fighting the directive, calling for a transparent and fair federal process grounded in technical realities rather than reactionary bans.
