Anthropic Recalls AI Models Due to Security Concerns

On Friday night, the AI company Anthropic abruptly disabled access to its most recent models, Fable 5 and Mythos 5, in response to a directive from the Trump administration. The order, issued by the US Commerce Department, was received just days after the models were launched. According to the company, this was the only way to ensure immediate compliance with the government's demand for export controls. The administration’s primary concern involves a reported jailbreak that allows users to bypass safeguards meant to block sensitive prompts. These prompts often relate to specialized fields like chemistry, biology, and cybersecurity. An official stated that the pause was requested so that the national security apparatus could be hardened against such threats, a process that might be completed within the next few weeks. Anthropic, however, has publicly disagreed with the necessity of this recall. They claimed that the government has only provided verbal evidence of a narrow, non-universal jailbreak. The company explained that this vulnerability involves reviewing a specific codebase for minor software flaws. Furthermore, they noted that other models, such as GPT-5.5, have similar capabilities. Anthropic warned that if this strict standard were applied to the entire industry, it would essentially halt the deployment of all new frontier models. This move follows an executive order signed by President Trump earlier this month, which urged AI developers to submit to voluntary security testing. While Anthropic has complied with the legal directive, they apologized to their users for the disruption and characterized the situation as a misunderstanding. The company plans to release more comprehensive details within 24 hours. While access to Fable 5 and Mythos 5 remains restricted, other Anthropic models continue to operate without interference.