Anthropic pulls Fable 5 and Mythos 5 offline after US export directive
Global access to Anthropic's latest AI models cut over nationality-based security concerns
Anthropic announced on June 13, 2026, that it had removed its latest large language models, Fable 5 and Mythos 5, from public availability. The move follows a directive from the U.S. government, which cited national security concerns and ordered Anthropic to restrict access for foreign nationals. Anthropic stated it could not reliably verify user nationalities on a global scale, so it chose to shut down access entirely—affecting users worldwide, including U.S. citizens.
The company publicly disagreed with the government's order, believing it was triggered by evidence of a potential narrow 'jailbreak' of Fable 5, which Anthropic argued was based on verbal reports and not robust testing. The incident highlights growing tensions between AI developers and export control regulations, with implications for model availability and international collaboration. Anthropic has not provided a timeline for reinstating access.
- Anthropic disabled Fable 5 and Mythos 5 globally on June 13, 2026.
- U.S. government directive cites national security, restricts foreign national access.
- Anthropic says the order stems from a verbal, narrow jailbreak claim it disputes.
Why It Matters
Regulatory friction forces global AI model shutdowns, challenging international access and developer autonomy.