SAN FRANCISCO — Anthropic on Tuesday released a formal compliance framework in response to the US government's order restricting international access to its most advanced Claude AI models, introducing tiered API access that limits capabilities available to users in designated non-allied nations. The move follows widespread criticism of Washington's directive, which sparked debate among researchers, civil liberties advocates, and enterprise customers who rely on Claude for global operations.

Under the new framework, Anthropic confirmed that its Claude 3.5 and Claude 4 series models will face usage restrictions in countries not covered by existing US technology export agreements. Developers and businesses in affected regions will be redirected to a reduced-capability API tier, while full model access remains available to verified customers operating within allied jurisdictions. Anthropic said the restrictions are designed to satisfy federal compliance requirements while minimising disruption to legitimate commercial users.

The announcement drew immediate reactions from the AI industry. Rivals including Google DeepMind and OpenAI were asked by analysts to clarify whether similar federal guidance had been extended to their own international API offerings. Neither company had issued a public statement by midday Tuesday, though industry sources suggested both firms were reviewing comparable compliance obligations stemming from the same executive order.

Anthropics chief policy officer issued a statement emphasising that the company had sought to balance national security requirements with its commitment to broad and equitable AI access. The statement acknowledged that the restrictions would affect thousands of developers in regions including parts of Southeast Asia, the Middle East, and Africa, and committed to a 90-day transition window to allow affected customers to migrate workflows or seek exemptions through a new verification portal.

Policy analysts in Washington noted that the order represents a significant escalation in the US government's effort to treat frontier AI models as dual-use technologies subject to export control regimes similar to those governing advanced semiconductors. The Anthropic compliance framework is expected to set a precedent that other frontier AI laboratories will be compelled to follow in the coming weeks, as federal agencies move toward formal rulemaking on AI export controls.