TL;DR: Anthropic took its newly released Claude Fable 5 and Mythos 5 models offline following a US export-control directive, highlighting growing government intervention in advanced AI safety and deployment.
Summary: Anthropic released Claude Fable 5, a public model with strict safety classifiers, and Claude Mythos 5, an unsafeguarded version restricted to security researchers. Just days later, the US government issued an export-control directive prohibiting foreign nationals from accessing either model due to concerns about guardrail jailbreaks. Anthropic subsequently complied by suspending access to both models, though they disputed the severity of the security risks cited by regulators.
Why it matters: This unprecedented regulatory action signals that developer access to advanced, dual-use models will face tightening government restrictions. Builders should monitor how these restrictions affect global API access and design fallback systems to handle sudden model deprecations.
Source: wired.com