In a dramatic escalating clash between the tech sector and the US government, Anthropic announced on Friday that it has suspended access to its two most powerful AI models, Fable 5 and Mythos 5. The abrupt move follows a sweeping national security directive issued by the US government.
The directive, which Anthropic received late Friday afternoon, explicitly bans all foreign nationals—including Anthropic’s own international employees—from accessing either model due to severe national security concerns.
The Core Dilemma: Advanced Capabilities vs. Cyber Vulnerabilities
The sudden enforcement action centers on the immense capabilities of Anthropic’s next-generation model family and fears over potential exploitation by hostile actors.
-
Mythos 5 (The Unreleased Frontier): This cutting-edge model has been held back from the general public, accessible only to select enterprise partners. Anthropic delayed its full launch due to internal concerns over its unprecedented ability to identify complex software vulnerabilities—flaws in code that hackers could exploit.
-
Fable 5 (The Rapid Takedown): Released to the public just three days prior on Tuesday, Fable 5 was engineered specifically as a “locked-down,” heavily safeguarded version of Mythos 5 designed for safe commercial use.
The “Jailbreak” Trigger
While the official directive did not detail the government’s exact technical reasoning, Anthropic understands that federal officials became aware of a narrow “jailbreak” method—a technique used to bypass an AI’s safety guardrails. The government fears this exploit could allow malicious actors to unlock Fable 5’s restricted code-analysis features and use them to launch cyberattacks.
Anthropic Pushes Back Against Federal Overreach
Anthropic has publicly voiced its disagreement with the suspension order. According to the company, its internal security testers analyzed the alleged exploit and found no evidence of a “universal jailbreak.”
The company warned that forcing a total commercial recall over a highly specific, narrow exploit sets a damaging precedent that could stifle the entire artificial intelligence sector.
“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, we believe it would essentially halt all new model deployments for all frontier model providers.”
— Anthropic Official Statement
Deepening Friction with the Trump Administration
The suspension is the latest flashpoint in an ongoing legal and ideological standoff between Anthropic and the Trump administration. The friction stems from a history of policy disagreements:
| Conflict Area | Anthropic Position | US Government Position |
| Military & Surveillance AI | Refused to allow its models to be adapted for mass surveillance or autonomous weaponry. | The Pentagon subsequently canceled multi-million dollar defense contracts with the firm. |
| Fable 5 Commercial Deployment | Argued the model is safe and its hacking utility doesn’t exceed existing public AI tools. | Commerce Secretary Howard Lutnick allegedly issued the urgent directive to freeze access. |
As of Friday evening, US Commerce officials have not responded to requests for comment regarding how long the suspension will last or if a specialized framework will be established to let domestic users regain access.
