The US Department of Commerce has lifted export restrictions on Anthropic's two most advanced AI models, Fable 5 and Mythos 5, after the company worked with federal authorities to address security vulnerabilities. Anthropic announced on July 1 that it would begin restoring global access to these models following the reversal of controls that had blocked their release since mid-June.

What Triggered the Export Ban in the First Place?

On June 12, US authorities abruptly ordered Anthropic to suspend access to Fable 5 and Mythos 5 for all foreign nationals after discovering vulnerabilities in the safeguards designed to prevent misuse of the technology. The move reflected growing federal concern that advanced AI models could be accessed by military or intelligence users in China, Russia, or other countries deemed a national security risk. The Trump administration had issued an executive order on June 2 calling for the federal government to establish a voluntary framework for private companies to test and release powerful "frontier" AI models in collaboration with government oversight.

The sudden lockdown caught the AI industry's attention, particularly because it signaled a shift toward more aggressive government vetting of AI releases. OpenAI, Anthropic's closest competitor, faced similar restrictions and delayed the full public launch of its GPT-5.6 model at the government's request, limiting access to a small group of vetted partners.

How Did Anthropic Resolve the Security Issues?

Anthropic worked closely with the Trump administration to strengthen its security protocols around the models. In a letter dated June 26, US Commerce Secretary Howard Lutnick confirmed that "Anthropic has worked with the US government to address risks associated with the Covered Models". Just four days before the full export ban was lifted, the company had already received authorization to allow a small group of American cybersecurity firms to access Mythos 5, signaling progress in negotiations.

The company's remediation efforts included three key commitments outlined in Lutnick's letter:

• Proactive Risk Detection: Anthropic agreed to proactively detect and address security risks associated with the models before they reach users.
• Government Coordination: The company committed to working diligently with the US government on protocols, standards, and release procedures for Mythos 5, Fable 5, and future models.
• Threat Reporting: Anthropic pledged to inform the US government of any malicious activity detected in connection with the models.

On July 1, the Trump administration formally withdrew the export restrictions, and Anthropic announced it would begin restoring access the following day.

What Does This Mean for the Broader AI Industry?

The lifting of the export ban signals that the US government is willing to work with AI companies to balance innovation with national security concerns, rather than imposing permanent restrictions. However, the process has drawn criticism from industry leaders. OpenAI CEO Sam Altman expressed concern about the government's role in selecting which companies can access powerful models. "This isn't quite the process that we think is optimal," Altman said on Friday, adding that while extensive safety testing "is not a bad idea," he dislikes "the idea of the government picking the customers".

Sam Altman

CIA Director John Ratcliffe defended the government's cautious approach by comparing advanced AI models to nuclear weapons. "In conversations with many of the president's other national security and economic security advisers, we're talking about the impact of these frontier AI models," Ratcliffe said during a speech at the AWS summit in Washington. "It would be not misplaced to refer to their capabilities as akin to digital nuclear weapons".

The episode underscores a fundamental tension in AI development: how to foster innovation while preventing misuse by hostile actors. Anthropic's successful negotiation with federal authorities suggests that companies willing to implement robust security measures and maintain transparent communication with government agencies can navigate these restrictions. Yet the process remains contentious, with industry leaders questioning whether government vetting of AI customers is the right approach.

Steps Companies Are Taking to Comply With Government AI Oversight

• Security Audits: Companies are conducting thorough security assessments of their models before release to identify and address vulnerabilities that could enable misuse.
• Government Liaison Teams: AI labs are establishing dedicated teams to work with federal agencies on protocols, standards, and release procedures for frontier models.
• Malicious Activity Monitoring: Companies are implementing systems to detect and report suspicious or harmful use of their models to relevant government authorities.

The resolution of Anthropic's export ban marks a turning point in how the US government and AI companies will interact going forward. Rather than permanent bans, the framework appears to favor conditional access based on demonstrated security improvements and ongoing collaboration. This approach may become the template for how other AI developers navigate federal oversight of their most powerful models.