The AI company suspended its newest and most advanced models for all customers to comply with a U.S. government order citing national security risks tied to a potential jailbreak.
Anthropic just pulled the plug on its most advanced AI models, Fable 5 and Mythos 5. The shutdown, which took effect yesterday, June 12, 2026, followed a direct order from the U.S. government citing national security concerns.
On June 12, Anthropic received a government order to suspend access to Fable 5 and Mythos 5 for all foreign nationals. This group includes individuals inside and outside the U.S., and even some of the company's own employees.
To comply with the broad directive, the company opted for a complete shutdown of the two models for all customers.
This shutdown is highly specific. All other Anthropic models, including its popular Claude chatbot, remain operational and unaffected. The company's announcement confirmed the move was made solely to adhere to the government's order.
The shutdown is significant because of the models involved. Fable 5 was only launched on June 9, just three days before the suspension. It was designed to bring the powerful capabilities of Mythos 5, Anthropic's elite cybersecurity model, to a wider audience with capabilities that reportedly "exceed" any previous model from the company.
While the government did not specify its concerns in the official order, Anthropic believes the action stems from a specific vulnerability. The company was informed that an unnamed entity shared verbal evidence with officials of a potential "narrow, non-universal jailbreak" for Fable 5.
A jailbreak is a method for bypassing an AI's safety features to make it perform tasks it was designed to refuse. This particular jailbreak, though described as narrow, was apparently sufficient to trigger government intervention.
Anthropic detailed its defense strategy, explaining that perfect resistance to jailbreaking is impossible for any model.
The company's approach focused on a two-pronged defense: make jailbreaks narrow or extremely difficult, and combine this with thorough monitoring to detect and shut down attacks quickly. The company also noted it had already instituted "strong safeguards" for Fable 5, so much so that many users had complained the restrictions were "overly broad."
In a firm public statement, Anthropic expressed its disagreement with the government's decision. The company argued that a potential jailbreak should not be grounds for recalling an entire commercial model.
Anthropic, a vocal proponent for more robust AI regulation, criticized the lack of a clear and fair process in this instance.
"As we have stated publicly, we believe the government should have the ability to block unsafe deployments, as part of a statutory process that is transparent, fair, clear, and grounded in technical facts," the company wrote. "This action does not adhere to those principles."
This shutdown is more than a technical issue; it's a clear line in the sand. As AI models capable of tackling sensitive cybersecurity tasks become public, the clash between developers and regulators is just getting started. Anthropic has promised to share more details within 24 hours, and the AI community is watching closely.
