"The US government’s experimentation with AI in rulemaking carries the risk that actions associated with the technology may be overturned in court, former federal officials warn.

Several agencies in President Donald Trump‘s second term are either using or testing AI to help with rulemaking, specifically around the management of the voluminous and burdensome public comment process. That development comes as Trump has pushed officials to roll out AI wherever possible, with his 2025 AI Action Plan calling for faster AI adoption across the government to “serve the public with far greater efficiency and effectiveness.”

The Centers for Disease Control and Prevention is using a generative AI tool to analyze public comments, judging whether a given commenter supports, opposes, or is neutral about a proposed rule. The Department of Transportation is using agentic AI to categorize comments and detect their sentiment, according to its AI use plans.

The challenge is that agencies are required under the Administrative Procedure Act to use reasoned decision-making during the rulemaking process. And it’s generally not possible to understand how most current AI systems reach their conclusions, according to Hans van Oostrom, director of the AI Academic Initiative Center at the University of Florida."

Stephen Lee reports for Bloomberg Environment August 13, 2026. [2]