US Lifts Export Restrictions on Anthropic's Advanced AI Models After Security Review

The US Department of Commerce announced on Wednesday the removal of export restrictions on advanced artificial intelligence models developed by Anthropic, including the popular Claude AI. The lifted restrictions apply to the company's latest Fable and Mythos models, according to Reuters. This decision follows a suspension less than three weeks ago, when Anthropic had to block access to its most advanced AI models due to national security concerns. The heightened scrutiny of advanced AI technologies stems from fears that countries like China and Russia could exploit them for military purposes.

US Secretary of Commerce Howard Lutnick stated that Anthropic committed to cooperating with the government to identify potential risks and report any malicious activity. In response, Anthropic announced it would restore access to the restricted models starting Thursday and continue coordinating with the government to expand access. According to Anthropic's website, the Fable model is designed with strong safeguards for general use, while Mythos has fewer restrictions and is intended for a trusted group of partners focused on cybersecurity. The company described Mythos as its most powerful model for cybersecurity and biological research.

Last month, Anthropic had blocked all users from accessing its most advanced AI models after receiving a government directive restricting foreign nationals' access due to national security concerns. The government did not provide specific details about these concerns but reportedly feared that the Fable 5 model's defenses could be breached to identify software vulnerabilities. This move reflects broader US efforts to regulate AI technology exports amid geopolitical tensions and security risks.