White House blocks Anthropic's Claude Mythos and Fable 5 over national security concerns
Tags AI · National Security · Export Controls
The US government forced Anthropic to pull its two newest AI models, Claude Mythos and Fable 5, just three days after release, citing national security concerns. The action followed claims that SK Telecom's access to Claude Mythos posed risks due to alleged ties to China, and that Amazon researchers found a way to bypass Fable 5's guardrails. Anthropic has been unable to redistribute the models since, and the White House has indicated that rerelease would require guaranteeing the models' guardrails cannot be circumvented — a condition security experts say is technically impossible. Cybersecurity researchers signed an open letter calling the move dangerous, noting the same jailbreaks exist in other widely available models. The episode marks the first time the US government has used export control authority to restrict access to a commercial AI model after public release.
Technical significance
This sets a precedent for post-release government intervention in AI model distribution. The requirement to 'block all jailbreaks' before rerelease is considered technically infeasible by security researchers, potentially creating a de facto permanent ban. This could accelerate the development of open-source alternatives and push AI labs to preemptively restrict capabilities to avoid regulatory entanglement.