Trump admin says Anthropic's 'recklessness' triggered export controls on latest AI models

The U.S. Commerce Department banned Anthropic’s Fable 5 and Mythos 5 AI models from use by foreign nationals, citing national security concerns. A senior Trump administration official told FOX Business that Anthropic’s response to early warnings about vulnerabilities in the models led to the export controls, describing the company’s actions as 'reckless.' The government claims Anthropic failed to address concerns seriously enough before the June 9 release, despite expectations set by the new AI executive order. The ban followed Amazon AI experts successfully 'jailbreaking' Fable 5, exposing full cyber capabilities in the model. Mythos 5, released to a limited group, was also restricted. Anthropic stated the models were designed to constrain cyberattacking abilities, but external tests revealed workarounds. Anthropic disputes the administration’s claims, arguing it engaged with regulators within 15 minutes of initial concerns. The company denied CEO Dario Amodei was unreachable during a wellness retreat, as suggested by the official. Sources close to Anthropic say the government did not provide specific details about the vulnerabilities, leaving the company unable to address them effectively. The Trump administration’s decision reflects an 'all-of-government' approach to AI safety, involving the Commerce and Treasury Departments. Officials view this as a test case for enforcing new AI guardrails, signaling that companies leading AI development must prioritize compliance. A tech policy expert described the situation as a standard government stance: if a company fails to address safety concerns, regulators will take a firm position. The export controls mark a significant escalation in U.S. oversight of AI development, raising questions about industry cooperation with government regulations. The dispute highlights tensions between rapid innovation and national security priorities in the AI sector.