US Government Restricts AI Model Access

The U.S. Commerce Department issued an order on June 12, compelling Anthropic to immediately restrict foreign national access to its advanced AI models, Claude Fable 5 and Mythos 5. In response, Anthropic disabled these models for all users within hours to comply with the directive, which cited national security concerns. The order barred access to noncitizen employees and all foreign nationals, creating an insurmountable challenge for Anthropic to verify user nationalities in real-time. The restrictions were lifted on June 30 after Anthropic implemented enhanced safety guardrails. However, these new guardrails significantly degraded the models' performance, causing benchmark scores to "collapse" and reducing their overall intelligence and capacity. One experiment indicated that Fable 5 could only complete 3 out of 12 routine tasks after the implementation of the new controls. The Commerce Department's action, taken under the Export Control Reform Act of 2018, has raised significant concerns within the U.S. AI industry regarding the government's ability to act as a gatekeeper for AI models and the legality of such interventions, particularly as the Act was primarily designed for hardware exports. The incident prompts fundamental questions about the government's authority to control access to AI technologies and the potential impact on innovation and global collaboration.