Anthropic Halts New AI Models Following US Government Security Directive
Anthropic has paused the deployment of its advanced artificial intelligence models, Claude Fable 5 and Mythos 5, after US national security authorities issued a directive citing security concerns. The company confirmed that the order specifically mandated the suspension of access for foreign nationals, necessitating a blanket disablement for all customers.
The company, which had previously described Fable 5 as 'too powerful', stated that authorities had not identified specific vulnerabilities. Instead, Anthropic indicated that the government believed a method for 'jailbreaking' Fable 5 had been discovered. Jailbreaking, in this context, refers to bypassing software restrictions to gain unauthorised access or features.
Anthropic acknowledged reviewing a demonstration of this technique, which revealed a 'small number of previously known, minor vulnerabilities'. The company contended that these issues appeared 'relatively simple' and that 'other publicly-available models' could also identify them without requiring a bypass method.
This development unfolds concurrently with an existing legal dispute between Anthropic and the Trump administration. The company is currently challenging a designation by then US Defence Secretary Pete Hegseth, who labelled Anthropic a 'supply chain risk'—a designation typically reserved for companies in adversarial nations, implying insufficient security for government use. A US judge has temporarily blocked the enforcement of this directive, allowing government entities to continue using Anthropic's tools while the lawsuit proceeds.
