Anthropic’s safety warnings may have just backfired — the government has pulled the plug on its most powerful AI
📰 ArticleConnie Loizos
US government orders Anthropic to shut down Claude Fable 5 and Mythos 5 over national security concerns.
The directive, received on June 12, 2026, forces Anthropic to disable both models globally. Mythos 5, previewed in April, finds flaws in all major OS and browsers and was shared with only 50 vetted organizations via Project Glasswing. Fable 5, released three days earlier, had guardrails. Anthropic argues the claimed jailbreak (prompting to read a codebase and identify flaws) is narrow and already available in other models like GPT-5.5. Anthropic fears halting all new model deployments if this standard is applied industry-wide.
💡Highlights
- ├─Claude Fable 5 & Mythos 5 shut down globally
- ├─Government cites national security & jailbreak concerns
- └─Anthropic claims narrow jailbreak already exists elsewhere
🎯For
- ├─AI community
- ├─Policy makers
- └─Tech industry