US Orders Anthropic to Disable Claude AI Models Over Jailbreak

Anthropic has disabled two recently launched AI models following a Friday afternoon directive from the US government that cited national security concerns, according to details first reported by WIRED.

The company removed access to Claude Fable 5 and Mythos 5 for all customers to comply with an export control order it received at 5:21 PM ET Friday. While the directive specifically targeted access by foreign nationals both inside and outside the United States—including Anthropic's own foreign national employees—the company chose a complete shutdown to ensure compliance.

The jailbreak concern

The government order stems from what officials believe is a method to bypass the safety guardrails built into Claude Fable 5. According to Anthropic, the administration provided only verbal evidence of a potential jailbreak technique that involves asking the model to read specific code and identify software vulnerabilities.

Anthropic reviewed a demonstration of the technique and found it identified only a small number of previously known, minor vulnerabilities. The company maintains these flaws are relatively simple and that other publicly available AI models can discover them without requiring any bypass method.

"The letter did not provide specific details of its national security concern," Anthropic wrote in a Friday blog post responding to the order.

Background on the models

Anthropic released Claude Fable 5 publicly on Tuesday as a safeguarded version of its Mythos AI model. The company built specific restrictions into Fable 5 to prevent it from answering questions about cybersecurity, biology, and chemistry. The underlying Mythos Preview model had a limited rollout in April, designed to give organizations access to powerful cybersecurity capabilities for defensive purposes while preventing exploitation by malicious actors.

Anthropic said it conducted the public release in collaboration with the US government.

Why it matters

This represents the first known instance of the US government ordering an AI company to disable models post-release based on national security concerns. The incident highlights the growing tension between AI companies' desire to deploy powerful models and government concerns about dual-use capabilities—particularly in cybersecurity. It also raises questions about the process for evaluating AI safety: Anthropic CEO Dario Amodei advocated earlier this week for a "fair, structured, and transparent government process" to block unsafe AI releases, yet the company argues this order failed to meet those standards.

Escalating tensions

The order arrives amid deteriorating relations between Anthropic and the Trump administration. Earlier this year, the Department of Defense designated Anthropic a "supply chain risk" after the company attempted to set restrictions on how the military could use Claude. That designation effectively prohibited government agencies and contractors from using Anthropic's technology, prompting the company to file lawsuits against the administration.

The White House and US Commerce Department did not immediately respond to requests for comment, according to WIRED, which first reported these details.