Anthropic Restores Advanced Claude Models Following Government Safety Review

Anthropic has restored global access to its most advanced AI models, Claude Fable 5 and Mythos 5, after the U.S. Commerce Department lifted export controls that had forced an abrupt three-week suspension.

The models were taken offline on June 12 following a government directive citing national security concerns, including a report from Amazon researchers demonstrating a method to bypass Fable 5's cybersecurity safeguards.

Commerce Secretary Howard Lutnick confirmed the reversal in a letter to Anthropic, stating the company had "taken steps in close coordination with the U.S. government to address the risks."

The restoration comes with significant new safeguards. Anthropic has deployed an improved safety classifier that blocks the reported bypass technique in more than 99 percent of cases, redirecting flagged requests to Claude Opus 4.8.

The company acknowledged this may occasionally block legitimate coding requests but described the "safety margin" as necessary to prevent misuse. Anthropic categorized all reported jailbreaks as "minor" and confirmed that no universal jailbreaks capable of unlocking broad harmful capabilities have been discovered.

"We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers." – Anthropic, in its initial statement on the suspension

Beyond the technical safeguards, Anthropic has committed to an expanded relationship with U.S. government agencies under the June 2 AI executive order signed by President Trump. 

This includes pre-release access for government testing of frontier models, faster information sharing on jailbreaks and misuse patterns, dedicated joint research teams, and participation in a new cybersecurity vulnerability clearinghouse.

The company is also working with Amazon, Microsoft, Google, and other Project Glasswing partners to develop a consensus framework for assessing AI jailbreak severity and appropriate developer responses.

The episode marks one of the most significant government interventions in AI deployment to date. The Trump administration's approach reflects a shift toward more direct oversight of frontier AI, with the Department of Commerce working closely with Anthropic to analyze and approve Fable 5 before lifting restrictions.

As Anthropic moves toward an anticipated IPO, the resolution suggests a willingness to work within government frameworks while preserving the ability to deploy cutting-edge models. The broader question remains: will this become the new normal for frontier AI releases?