Anthropic abruptly pulled access to two of its newest and most capable AI models — Fable 5 and Mythos 5 — over the weekend, following intervention by the U.S. government, according to CNET. The federal restrictions centered on sweeping new limits over who is permitted to use the models, tied to claims that the systems could be jailbroken.
The move comes amid broader questions about whether Anthropic's latest Claude models can effectively be kept under human control, a concern flagged by AI Magazine. The sudden withdrawal surprised users and businesses who had only recently gained access to the models, with Memeburn noting that Fable 5 had been pitched as bringing Mythos-level AI capabilities to a wider audience.
Fortune framed the episode as a significant corporate governance test, arguing it exposed a crisis that every CEO in the AI industry should be paying attention to. The publication suggested the situation demands a clearer framework for how companies respond when regulators suddenly challenge the safety or accessibility of flagship products.
Anthropics' decision to comply quickly with federal demands — pulling the models rather than contesting the restrictions publicly — signals how much leverage U.S. regulators now hold over frontier AI labs, even as those labs race to release ever more powerful systems.
The episode matters because it shows that cutting-edge AI models can be pulled from the market almost overnight, and that government concerns about jailbreaking — the ability to get AI to bypass its own safety rules — are now consequential enough to halt product rollouts entirely.