Anthropic’s Most Powerful AI Models Are Back — Here’s What Changed

TLDR

• The U.S. lifted export controls on Anthropic’s Mythos 5 and Fable 5 models on June 30
• Controls were imposed June 12 after a jailbreak was found that exposed software vulnerabilities
• Anthropic suspended access for all users as it couldn’t verify nationalities in real time
• Anthropic agreed to share threat intelligence and give government agencies early model access
• Competition from Asian AI models like Fugu and Tulongfeng helped push the U.S. to ease restrictions

---

Anthropic is restoring public access to its two most advanced AI models after the U.S. government removed export controls that had blocked them for most of June.

Claude Fable 5 will be available again globally tomorrow.

After a series of productive conversations with the US government, we’re redeploying the model with a new set of classifiers to target and block more cybersecurity tasks. In the near term, some routine tasks like coding…

— Anthropic (@AnthropicAI) July 1, 2026

Fable 5 returns globally on July 1 across Anthropic’s platforms. Mythos 5, which shares the same underlying model but has fewer safety restrictions, is being restored to a limited group of U.S. organizations that received government approval on June 26.

Why the Models Were Blocked

The restrictions began on June 12 when the government added both models to its list of export-controlled technologies. This meant they could not be made available to foreign nationals without special approval.

The trigger was a cybersecurity finding. Amazon researchers discovered a jailbreak — a method to bypass the model’s safeguards — that caused Fable 5 to identify software vulnerabilities and, in one case, produce code showing how one could be exploited.

Because the order took effect immediately and Anthropic had no way to verify users’ nationalities in real time, the company suspended access for everyone rather than risk breaching the rule.

Cybersecurity experts were skeptical of the restrictions from the start. They noted that Anthropic had already pledged to address many of these security concerns voluntarily, months before the export rule existed. Some analysts suggested the ban functioned more as leverage than a genuine security measure, linked to Anthropic executives’ public criticism of the Trump administration’s approach to AI.

What Anthropic Agreed To

Secretary of Commerce Howard Lutnick said Anthropic agreed to proactively detect and address security risks, work with the government on protocols for future model releases, and inform U.S. authorities of any malicious activity.

Anthropic is also deepening its collaboration with government agencies. This includes giving designated agencies early access to frontier models before public release and sharing threat intelligence.

The company is drafting a framework with Amazon, Microsoft, and Google to score how dangerous a given jailbreak is. These arrangements tie into a June 2 executive order on AI security.

Global Competition Played a Role

Asian AI companies have begun releasing models approaching Mythos-level capabilities, including Fugu and Tulongfeng. That competitive pressure put the U.S. government in a difficult position, making it harder to justify keeping American AI models off global markets.

Last week, Lutnick also cleared Mythos to be released to select White House-approved customers. OpenAI’s latest models were distributed in a similar way, going to a group of approved organizations rather than the general public.

Anthropic is privately held, but traders have bet on its value through a pre-IPO futures contract on the onchain exchange Hyperliquid. That contract fell about 3.7% when access was suspended in June.

The Trump administration’s shifting AI policy has left many in the industry uncertain about what rules will govern future model releases.