Anthropic AI models faced US export controls and here is why they returned

Anthropic has officially reopened global access to its advanced Claude Fable 5 and Mythos 5 AI models, marking a significant win for the tech sector after a brief regulatory pause. 

The US government lifted its export restrictions once the company implemented comprehensive new safety measures. These restrictions, imposed early last month, temporarily halted international access while federal agencies assessed the models’ security features.

US Commerce Secretary Howard Lutnick confirmed the positive outcome, emphasizing that close cooperation between the government and Silicon Valley ensures safe deployment without hindering innovation.

“Over the past two weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US government and strengthen America’s leadership in AI,” Lutnick stated on social media.

Following approval, Anthropic swiftly restarted access for all users worldwide. The company is restoring functionality across major platforms such as AWS, Google Cloud, and Microsoft Foundry, enabling developers to immediately utilize the models for complex coding and knowledge tasks.

Additionally, Anthropic is progressing with Mythos 5, a specialized version tailored for cybersecurity defense. Through its Project Glasswing, the firm is securely providing Mythos 5 to a select group of trusted US organizations, with plans to expand access to more domestic and international partners soon.

To obtain government clearance, Anthropic incorporated advanced safety classifiers into the models. These AI systems automatically detect and prevent requests intended to exploit the models for malicious cybersecurity purposes.

READ MORE: Anthropic gets US approval to restore powerful AI models

Collaborating closely with federal agencies, Anthropic demonstrated that these safeguards can prevent misuse while maintaining the models’ utility for routine enterprise functions.

Industry experts praised the swift resolution. Cybersecurity specialists stressed that reintroducing these models is vital for developing defensive software.

Keven Knight, CEO of Talion, explained that restricting access would disadvantage Western organizations amid the rapid growth of global AI development. Mike Britton, CIO of Abnormal AI, emphasized that the controlled deployment of Mythos 5 to trusted institutions is the right approach for ensuring genuine security.

He noted that creating more secure software depends on disciplined access controls over time.