How the Trump administration shifted from a “light-touch” approach to AI policy to an interventionist stance that led to restrictions on top AI models in the US
In January 2025, just three days after entering office for his second term, President Donald Trump signed …
Context & Ripple Effects
The administration’s posture had already been moving away from broad deregulation: its AI Action Plan sought to roll back selected Biden-era safeguards, while subsequent coverage described case-by-case interventions without clear rules. Discussion of a working group to examine model vetting signaled that access to advanced systems could become a policy lever rather than solely an industry decision.
First-order effects
- Developers of the affected leading AI models must operate under new US restrictions, making model availability and release decisions subject to federal intervention.
- Business users and downstream deployers of restricted models face less certainty about which top-tier systems they can access in the US.
Second-order effects
- Rival model providers gain an incentive to adapt products, distribution, and compliance plans to shifting government preferences rather than a stable, general framework.
- The absence of clear, uniform rules raises planning costs across the AI supply chain, because firms must price in the possibility of further model-specific decisions.
Third-order effects
- If this approach persists, US AI governance could evolve toward discretionary frontier-model access control—less like a centralized regulator and more like industrial policy administered through individual interventions.
- That would make the commercial value of frontier models increasingly dependent on alignment with national policy priorities, though the durability of this model depends on whether the administration codifies clearer standards.
The trend: US AI policy is shifting from a light-touch posture toward state-directed control over access to frontier models.