Around June 12, 2026, the US Commerce Department under the Trump administration issued export controls ordering an immediate halt to access to Anthropic's latest frontier models, "Claude Fable 5" and "Claude Mythos 5," by foreign nationals. Citing the difficulty of nationality screening, Anthropic temporarily disabled both models for all customers, including those in the US.
June 12, 2026 · US Commerce Department / Anthropic
US Export Controls Force Anthropic to Pull Its Newest Frontier Models — Worldwide
Ordered to block foreign nationals from accessing Claude Fable 5 and Mythos 5 on national-security grounds, Anthropic said reliable nationality screening was impossible — and disabled both models for all customers, including those in the US.
2
frontier models disabled — Fable 5 & Mythos 5
~3
days from public release to full shutdown
ALL
customers cut off, not just foreign nationals
A four-day timeline, from launch to lockout
JUN 9–10
Fable 5 released to the public; Mythos 5 offered to approved orgs via Project Glasswing
→
JUN 12
Commerce Dept. (BIS) orders immediate halt to foreign-national access on security grounds
→
SAME EVENING
Anthropic disables both models for everyone; calls the action "a misunderstanding"
CLAUDE FABLE 5
First public Mythos-class release
Safeguards limit dangerous capabilities. Available on Amazon Bedrock, Google Cloud and the Anthropic platform.
CLAUDE MYTHOS 5
Upper-tier, Fable-equivalent capability
Limited access for approved organizations via Project Glasswing only.
The stated concern
The directive cited national security but did not specify the precise concerns.
Jailbreak techniques could enable misuse for cyberattacks or hacking
Reported concern over access by China-linked entities
Nationality screening deemed impossible — so all users were blocked
SUPPORTERS / OPTIMISTS
Administration stresses keeping AI leadership in the US
Frames it as rapid response to a national-security risk
Developers expect gains in software engineering and long-running tasks
CRITICS / SKEPTICS
Move criticized as ad hoc and capricious
Could weaken US firms and hand rivals an advantage
Some see Anthropic's blanket shutdown as an overreaction
What's next: with the jailbreak details undisclosed and many users forced onto models like Opus 4.8, attention turns to whether this becomes a precedent for future AI export controls.
Continue reading The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in ✓ Signed in — this article isn’t included in your current plan.Unlocking the full article…