US government staggers GPT-5.6, suspends Fable 5

This unprecedented federal intervention marks the end of self-regulation for frontier AI labs, showing that national security concerns will now dictate release schedules. Developers must prepare for a fragmented API landscape where access to cutting-edge models is gated by government vetting.

• –Anthropic’s global suspension of Fable 5 and Mythos 5, triggered by export control compliance issues and jailbreak concerns, highlights the difficulty of policing access by nationality.
• –The customer-by-customer approval process for OpenAI's GPT-5.6 limited preview sets a restrictive precedent that could delay integration and deploy cycles for enterprise users.
• –Under-the-hood red-teaming under Project Glasswing reportedly alarmed officials by showing Mythos finding critical infrastructure vulnerabilities within hours, accelerating the regulatory crackdown.
• –This aggressive oversight risks driving open-weights development and deployment overseas to avoid compliance bottlenecks in the United States.