President Trump signed a scaled-back AI cybersecurity executive order establishing a framework for federal evaluation of artificial intelligence models. The directive tasks the Office of the National Cyber Director with developing a process to share vulnerability information with critical infrastructure operators before public release. Trump had postponed the announcement originally scheduled for May 21 after expressing dissatisfaction with certain aspects.

The revised order requests some AI companies to submit their most powerful models for voluntary government review 30 days before public availability. This represents a compromise from earlier drafts that proposed 90-day review periods, which industry officials reportedly wanted reduced to as little as 14 days. The administration had previously avoided substantive AI regulation, focusing instead on ideological restrictions.

Samir Jain from the Center for Democracy and Technology criticized the order as "opaque," noting it lacks transparency in benchmarking processes. Trump's decision to regulate AI security marks a departure from his administration's historically laissez-faire approach to AI risks. Previous regulations targeted ideological concerns rather than potential security vulnerabilities.