Trump Administration Establishes Voluntary AI Model Vetting Framework for National Security

Summary

President Trump signed an executive order on Tuesday establishing a federal framework to vet the national security risks of advanced AI systems before their public release. The order applies to 'frontier labs'—companies like Anthropic, OpenAI, and Google—developing the most advanced AI systems. Participation in the vetting process is voluntary, and the government must complete security reviews within 30 days of an AI model announcement.

The executive order represents a compromise after Trump initially postponed signing a similar policy on May 21, citing concerns that government oversight could harm America's technological lead against China. The current version emphasizes a narrower scope and shorter timeline, focusing specifically on cybersecurity and critical infrastructure protection rather than broad AI governance. The White House stated it is "NOT conducting oversight of all new models, as that level of government overreach would have chilling effects on free speech and innovation."

However, policy analysts have raised concerns about the vagueness of selection criteria and the concentration of discretion in the NSA director's hands. Some worry the policy could enable selective enforcement against specific companies. The timing follows Anthropic's April announcement of Claude Mythos, its most advanced AI model, amid an ongoing legal dispute between Anthropic and the Trump administration over a Pentagon contract.

• Critics warn the policy lacks clear criteria for model selection and grants significant discretion to the NSA director, raising concerns about potential selective enforcement against competitors