Policy

Anthropic Disables Fable 5 and Mythos 5 Worldwide After U.S. Export-Control Directive

On June 12, 2026, Anthropic received a U.S. government export-control directive at 5:21 p.m. ET ordering it to suspend access to Claude Fable 5 and Mythos 5 for any foreign national—inside or outside the United States, including foreign-national Anthropic employees. Unable to enforce citizenship filters in real time, Anthropic disabled both models for all customers the same evening. The company disputed that a narrow alleged jailbreak justified the recall, compared the demonstrated capability to publicly available models including GPT-5.5, and said other Claude models remained available.

Tech Insights Reporter Jun 12, 2026 6 min read San Francisco, CA

TLDR

Three days after launch, Anthropic turned off Claude Fable 5 and Mythos 5 for every customer on June 12, 2026 to comply with a U.S. export-control directive. The order—received at 5:21 p.m. ET and attributed to Commerce/BIS under national-security authorities—barred access by any foreign national worldwide, including Anthropic’s own non-U.S. staff. Anthropic argued the government’s concern appeared to rest on a narrow, non-universal jailbreak demo of minor, already-known vulnerabilities that other public models (including GPT-5.5) can also surface, and said it was working to restore access while complying.

What the directive required

From Anthropic’s primary statement:

Scope: Suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the U.S., including foreign national Anthropic employees.
Compliance choice: Because real-time citizenship gating across a global product is not reliable, Anthropic disabled both models for all customers.
Unaffected: Access to all other Anthropic models continued.
Timing: Directive received 5:21 p.m. ET on June 12; models had launched June 9.

Reporting (NYT, CNBC, Axios) attributed the letter to the Commerce Department (BIS), with some accounts describing ~90-minute compliance pressure and prior industry alerts to government about model capabilities.

Anthropic’s rebuttal

Anthropic’s statement emphasized:

Extensive pre-launch red-teaming with U.S. government, UK AISI, and third parties; no universal jailbreak found in thousands of hours of testing.
Safeguards deliberately conservative (users already complained of over-blocking).
Government evidence (as understood by Anthropic) showed a narrow technique—essentially reading a codebase and fixing flaws—without Mythos-specific uplift, comparable to everyday defensive use of other frontier models.
The company supports statutory, transparent government power to block unsafe deployments—but argued this action lacked that process and, if generalized, would halt frontier deployments industry-wide.
Framing: misunderstanding; working to restore access ASAP.

Why this story matters

This is the first time a newly launched frontier commercial model family was globally killed within days by a citizenship-based export-control order. It forces a hard collision among product launch velocity, dual-use cyber risk, immigration status of staff and users, and ad-hoc national-security process. Whether the episode is a one-off “misunderstanding” or a template for future model freezes will shape how labs stage Mythos-class releases and how allies view U.S. control over AI stacks they already depend on.

Sources

Anthropic: “Statement on the US government directive to suspend access to Fable 5 and Mythos 5” (anthropic.com/news/fable-mythos-access, June 12, 2026). Primary company statement.
CNBC, NYT, Reuters, Axios, Bloomberg same-night/next-day reporting on Commerce/BIS directive and global disable (June 12–13, 2026).
Anthropic launch post for Fable 5 / Mythos 5 (June 9, 2026) for timeline context.

Featured Image Alt Text

Claude Fable 5 and Mythos 5 logos grayed out behind a U.S. export-control stamp and offline indicator.