Introduction

The United States announced on 1 July 2026 that it has removed export‑control restrictions on Anthropic’s advanced large‑language models Claude Fable 5 and Mythos 5. The decision ends a three‑week shutdown that halted access for foreign nationals and forced the company to suspend the models across all its platforms.

Impact of the Ban

Export controls were imposed on 12 June, just three days after the models launched, citing national‑security concerns over a potential “jailbreak” that could bypass built‑in safeguards. Anthropic responded by blocking access for all foreign users, including its own non‑U.S. employees, and took the models offline on Claude.ai, its API, AWS Bedrock, and partner services. The rapid segmentation required proved technically unfeasible, creating operational chaos for customers worldwide.

Lifting the Controls

On 30 June the Department of Commerce notified Anthropic that the controls had been lifted, allowing the company to begin restoring access on 1 July. Commerce Secretary Howard Lutnick publicly noted that his team had worked closely with Anthropic to analyze and approve Fable 5, emphasizing “proactive detection and address[ing] security risks” and reaffirming America’s leadership in AI.

Implications for AI Policy

The reversal highlights a maturing regulatory framework for frontier AI. While the initial ban underscored uncertainty around advanced models that can discover software vulnerabilities, the swift resolution demonstrates that collaborative engagement between industry and government can produce balanced outcomes. Anthropic’s commitment to inform the government of any malicious activity and to enhance safeguards aligns with broader U.S. goals of promoting innovation while protecting national security.

Conclusion

With export controls lifted, Claude Fable 5 and Mythos 5 are set to resume global availability, restoring critical capabilities for developers, enterprises, and security researchers. The episode illustrates both the challenges and the evolving partnership needed to steward powerful AI technologies responsibly.