Anthropic says US government forced it to disable cybersecurity AI models

Anthropic disabled its Fable 5 and Mythos 5 models following a U.S. government export-control directive that bars foreign nationals from accessing the models, citing a reported method to bypass model safeguards; Anthropic says the issue is a minor, previously known jailbreak reproducible on other models and is contesting the government action as disproportionate and nontransparent.