The US has relaxed export controls on Anthropic’s AI models, allowing public access again. These models were previously restricted due to security concerns which proved impractical at scale.

Howard Lutnick, Secretary of Commerce, outlined new conditions for Anthropic including proactive detection and addressing of security risks, working with the U.S. government on protocols, and reporting malicious activity. These measures had already been publicly pledged by Anthropic months earlier, raising skepticism among cybersecurity experts who viewed the restrictions as leverage rather than a genuine security fix.

With competing AI models from Asian companies like Fugu and Tulonfeng, the US felt pressure to ease regulations on Anthropic to maintain global competitiveness. Last week, Mythos was released to select customers approved by the White House, mirroring similar moves for OpenAI’s latest models.

This shift highlights the Trump administration's inconsistent approach to AI policy, leaving companies unsure about future guidelines. An executive order issued in June signaled a desire to review models before release but faced criticism from influential analysts.