Donald Trump no longer considers Anthropic a threat to U.S. national security. A week earlier, the situation was different. The conflict centered on foreign access to the AI models Fable 5 and Mythos 5, which were temporarily blocked for all users following a Trump administration export control directive.
In an interview with Axios published last Friday, Trump said he might have viewed Anthropic as a threat. “Well, not now, but a week ago, maybe,” he replied when asked if he considered Anthropic or CEO Dario Amodei a security risk.
The tension stemmed from an export control directive issued by the Trump administration. The directive ordered Anthropic to block foreign users’ access to its most advanced models, Fable 5 and Mythos 5. However, Anthropic did not have a foolproof system to grant access only to Americans and therefore completely disabled both models.
Three words triggered the export ban
The three words “fix this code” proved sufficient to trigger the export ban. The Trump administration interpreted a test prompt regarding vulnerable code as a potential jailbreak that could expose security risks. More than a hundred security experts lodged a protest, arguing that the export ban hinders defenders outside the U.S., while attackers can switch to open-source or Chinese models that fall outside export controls.
Technical staff at Anthropic had previously met with Trump administration officials to discuss the dispute. Trump noted that CEO Dario Amodei had responded to the directive “very quickly” and “responsibly.”
Identity verification as the next step
Meanwhile, Anthropic is working on measures to comply with future export restrictions. The company is updating its privacy policy and has informed users in an email that they will need to verify their identity when using Claude Free, Pro, and Max in certain cases.
Trump did not rule out the use of emergency measures under the Defense Production Act, but also said, “But I’m not sure I have to do that.”
A spokesperson for Anthropic responded to the interview: “We are grateful to the administration for their ongoing partnership in working to get this matter resolved as quickly as possible. We remain committed to working alongside them towards our shared goals of protecting critical infrastructure and making sure the U.S. leads in AI.”