Anthropic, an AI company headquartered in San Francisco, is responsible for two extremely powerful AI models: Claude Fable 5 and Claude Mythos 5. Mythos 5 is Anthropic's most powerful model, which they have been very verbal about, and had to be heavily restricted and controlled to launch as Claude Fable 5 for public use. Recently, the U.S. government ordered them to shut off access to these AI models due to an alleged, narrow jailbreak where they will identify software flaws if given a codebase. Although Anthropic has complied to these orders, they have been open about their frustrations that a narrow jailbreak should not be enough to recall "a commercial model deployed to hundreds of people."