Anthropic has disabled two of its latest AI models, Claude Fable 5 and Mythos 5, after receiving an export control directive from the US government on Friday afternoon, according to a report by WIRED. The order, which cites national security concerns, requires Anthropic to suspend access to the models for any foreign national, including employees. In response, Anthropic removed access for all customers to ensure compliance.
Government Directive Without Specifics
Anthropic stated in a blog post that it received a letter from the US government at 5:21pm ET, which "did not provide specific details of its national security concern," according to WIRED. The company believes the government identified a method to bypass, or "jailbreak," Claude Fable 5. Anthropic reviewed a demonstration of this technique and found it identified a small number of previously known, minor vulnerabilities. The company argued that other publicly available models can discover the same vulnerabilities without needing a bypass.
"The letter did not provide specific details of its national security concern," Anthropic wrote.
Anthropic emphasized that it has implemented strong safeguards to reduce misuse and that the jailbreak is narrow and would not make an attacker meaningfully more dangerous than using another AI model. "To date, the government has only given us verbal evidence of a potential narrow, non-universal jailbreak, which essentially consists of asking the model to read a specific codebase and fix any software flaws," the company said.
Background of Conflict with Trump Administration
Earlier this year, the Department of Defense labeled Anthropic a "supply chain risk" after the company sought to impose restrictions on how the US military could use its technology, according to WIRED. That designation barred government agencies and contractors from using Anthropic's technology. In response, Anthropic filed lawsuits against the Trump administration.
The recent release of Claude Fable 5 and Mythos 5 on Tuesday was touted as a collaboration with the US government. Prior to public release, the Mythos Preview AI model had a limited rollout in April. The goal was to enable organizations to use its cybersecurity capabilities to improve defenses while mitigating risks of exploitation by bad actors to develop hacking tools.
Sources:
