Trump administration reignites its feud with Anthropic over latest AI models

The Trump administration escalated its dispute with Anthropic on Friday by ordering the AI company to block foreign access to its newest models, Fable 5 and Mythos 5. Anthropic complied, disabling access for all customers, though it stated the directive lacked explanation for the national security concerns. The company attributed the move to claims of a jailbreak method bypassing security restrictions, which it called overstated. The conflict stems from months of clashes over Anthropic’s AI use in military and intelligence applications, culminating in the Pentagon labeling the firm a supply-chain risk. Defense Secretary Pete Hegseth reinforced the administration’s stance on X, claiming the Pentagon had permanently barred Anthropic from its facilities, while David Sacks, a former Trump AI advisor, accused CEO Dario Amodei of refusing to address the jailbreak issue after a trusted partner raised concerns. Sacks claimed the administration had asked Amodei to resolve the problem but that he declined, though an unnamed source disputed this, saying Anthropic was willing to discuss the issue. Amazon reportedly played a key role in alerting the White House about security risks, though officials later called the company’s documentation misleading, as similar capabilities exist in OpenAI’s GPT-5.5. Anthropic’s decision came as the company prepared for an initial public offering, following a funding round valuing it at nearly $1 trillion. The move has reignited debates over government oversight of AI development, with critics questioning the administration’s abrupt intervention and the company’s transparency regarding security risks.