Trump Executive Order Empowers Federal Government to Pick Winners in AI Model Access
Key Takeaways
- ▸Trump's executive order creates a 30-day federal security review period for 'covered frontier models' before public release
- ▸The government will have authority to select which companies receive 'trusted partner' early access to advanced AI systems
- ▸Policy experts warn the provision could be weaponized against politically disfavored companies
Summary
President Trump has signed an executive order addressing advanced AI cybersecurity and frontier model deployment. While much of the order focuses on legitimate security initiatives—including establishing an 'AI cybersecurity clearinghouse' and directing federal grant programs toward AI vulnerability detection—the most controversial provision gives the federal government authority to review frontier AI models before public release and determine which companies receive early access to these systems.
Section 3 of the order ('Secure Frontier Model Deployment') requires AI companies to submit advanced models for up to 30 days of federal security review before release to 'other trusted partners.' More problematically, it directs AI companies to 'collaborate with the Federal Government to select trusted partners,' effectively allowing the administration to determine which companies gain early access to programs like Anthropic's Project Glasswing for Claude.
Policy experts have raised serious concerns that the government's discretion in selecting 'trusted partners' could be weaponized against companies perceived as political adversaries. Cato Institute analyst Juan Londoño warned that 'the government's involvement in decisions about which trusted partners can access these advanced models gives the executive a great deal of discretion' and 'could open the door to potential weaponization against companies that have any sort of conflict with the administration.'
The order was signed Tuesday following a postponement from May, with the review period reduced from 90 days to 30 days. While policy experts acknowledge the order addresses gaps in oversight, many are calling for Congress to establish clearer statutory rules that would constrain executive discretion and prevent politicized decision-making.
- The order also establishes an 'AI cybersecurity clearinghouse' and expands federal grant programs for AI vulnerability detection
- Review period was reduced from 90 days in May draft; experts call for Congress to establish clearer statutory guidelines
