This unprecedented incident marks the latest confrontation between Anthropic and the Trump administration. While the company says the order asked it to suspend access to “any foreign national, whether inside or outside the United States, including foreign national Anthropic employees,” it has removed access for all of its customers to ensure compliance.
Earlier this year, Trump’s Defense Department branded Anthropic a “supply chain risk” after the cloud-maker sought to draw red lines on how the US military could use its technology. The designation effectively barred government agencies and contractors from using Anthropic’s technology. Anthropic responded by filing a lawsuit against the Trump administration.
On Tuesday, Anthropic publicly released Cloud Fable 5, a version of the company’s Mythos AI model with safeguards that prevent it from answering questions about cybersecurity, biology and chemistry. Prior to the public release, which Anthropic conducted in collaboration with the US government, there was a limited rollout of Mythos Preview AI models in April. Its goal was to give companies and organizations the opportunity to use its powerful cybersecurity capabilities to improve their security, and to address concerns that the technology could be exploited by bad actors to develop powerful hacking tools.
In a blog post on Friday, Anthropic says it received a letter from the US government at 5:21 pm ET. “The letter does not provide specific details of his national security concern,” Anthropic wrote.
“Our understanding is that the government believes it has become aware of a method to bypass or ‘jailbreak’ Fable 5,” the company said. “We reviewed a demonstration of this specific technique being used to identify a small number of previously known vulnerabilities. All of these vulnerabilities appear to be relatively simple, and we have found that other publicly available models are also able to find them without the need for bypass.”
In the blog post, the company argued that it has implemented strong security measures to reduce the potential for misuse of CloudFable 5. Anthropic also claimed that the jailbreak the US government found for CloudFable 5 was narrow, and could not make an attacker meaningfully more dangerous than any other AI model.
“To date, the government has given us only verbal evidence of a potentially narrow, non-universal jailbreak, which essentially asks the model to read a specific codebase and fix any software flaws,” the company said in its blog post. “Our understanding is that a potential jailbreak was shared with the government.”
Spokespeople for the White House and the U.S. Commerce Department did not immediately respond to WIRED’s requests for comment.
Anthropic CEO Dario Amodei said in a policy essay earlier this week that he and the company support a fair, structured and transparent government process that would prevent the release of unsafe AI models. In a company blog post Friday, Anthropic argued that “this action does not comply with those principles.”
<a href