Fable 5 disabled following government directive

Anthropic abruptly shut down its newly launched Fable 5 and Mythos 5 AI models after receiving a directive from the US Commerce Department imposing immediate export‑control restrictions. The order followed concerns inside the Trump administration about a reported jailbreak that could bypass the models’ safety classifiers and assist with cybersecurity‑related tasks, potentially posing a national‑security risk. Anthropic says the government only provided verbal evidence of a narrow, non‑universal jailbreak involving code‑review capabilities, and argues that similar abilities exist in other models like GPT‑5.5. Still, the company says it must comply with the legal directive, disabling both models for all users worldwide. The administration reportedly wants time to “harden” national‑security systems before allowing the models to return.

Read the full story on Ars Technica →