Updated
Updated · The New York Times · Jul 13
Anthropic Restored Claude Fable 5 by July 1 After Trump Ban Hit Non-U.S. Users
Updated
Updated · The New York Times · Jul 13

Anthropic Restored Claude Fable 5 by July 1 After Trump Ban Hit Non-U.S. Users

3 articles · Updated · The New York Times · Jul 13

Summary

  • Claude Fable 5 returned online by July 1 after the Trump administration lifted restrictions that had blocked Anthropic’s newly released coding-focused model.
  • The government had banned the model for non-U.S. citizens days after its June 9 launch, citing safety concerns about the more powerful system.
  • Anthropic then shut off access for everyone because it could not reliably identify users’ nationality and reportedly had only 90 minutes to comply.
  • Officials later said they were satisfied with Anthropic’s steps to detect and address safety risks, ending a dispute that has fueled broader debate over AI regulation and free speech.

Insights

What precedent does a 90-minute takedown order set for the future of global AI development?
How will new AI export controls reshape the global balance of technological power?

Fable 5 Suspended: How a U.S. Ban, Jailbreak Risks, and Global Competition Reshaped Anthropic’s AI Market in 2026

Overview

Following the Commerce Department’s decision to lift export controls on June 30, 2026, Anthropic’s Claude Fable 5 and Mythos 5 quickly returned to general availability, alongside the launch of Claude Sonnet 5. The temporary suspension of Fable 5 exposed a major industry gap: there was no standard way to identify or score 'jailbreaks'—methods to bypass AI safeguards. In response, Anthropic began leading a collaborative effort with major tech partners to create a comprehensive standard for evaluating jailbreaks, focusing on capability gain, breadth, ease of weaponization, and discoverability. This marks a significant step toward stronger, industry-wide AI safety practices.

...