In brief
Anthropic halted the release of its powerful Claude Fable 5 model after U.S. officials flagged possible security risks. The company was instructed to block foreign users from the tool, citing concerns over a jailbreak method that could expose vulnerabilities. A lawsuit between Anthropic and the U.S. Pentagon continues amid the Trump administration's scrutiny. The move underscores growing worries about AI tools' safety for government use.
Anthropic, a U.S.‑based artificial‑intelligence company, has pulled the powerful Claude Fable 5 model after a brief period of public availability, following an order from U.S. authorities.
The order came three days after the tool’s unveiling. According to a statement on Anthropic’s website, federal regulators instructed the company to disable Fable 5 (and the related Mythos 5) for foreign nationals to comply with national‑security guidelines.
The regulators warned that a jailbreak technique could bypass the model’s safety safeguards and expose previously known vulnerabilities. Anthropic said the vulnerabilities were relatively simple and that other publicly available models could detect them without a jailbreak.
The suspension occurs amid an ongoing lawsuit between Anthropic and the Pentagon, in which the defense department has dubbed the company a “supply‑chain risk.” A judge has so far ruled the Pentagon’s directive cannot be enforced, allowing government agencies to keep using the tool while the dispute proceeds.
The episode highlights the growing tension between rapid AI innovation and the need for robust security checks, especially for tools that could have national‑security implications.
