EPA Deploys AI for Chemical Safety Without Public Notice

The Environmental Protection Agency is preparing to use custom artificial intelligence tools to speed up the approval of new toxic chemicals and pesticides. According to a review by GN technics/ai (en-US), the agency has failed to disclose these plans to the public, a move that legal experts describe as likely unlawful. Federal law requires the EPA to maintain a public registry of AI uses so that citizens can assess potential risks to privacy and safety before these systems are deployed.

This initiative aligns with a broader agency goal to position the United States as a leader in AI adoption. However, critics argue that skipping the transparency step removes a critical check on government power. By keeping the tools hidden, the agency prevents the public from understanding how decisions about dangerous substances are made, raising concerns about inherent biases and a lack of accountability in regulatory processes.

Legal mandates require public oversight

Congress established rules requiring federal agencies to list current and planned AI applications in an online directory. This system allows for public feedback and ensures that the government protects civil rights and public safety. The EPA’s recent registry does not include the new chemical and pesticide vetting tools, creating a gap between legal requirements and agency practice. Environmental advocates warn that this lack of transparency leaves taxpayers in the dark about how potentially hazardous materials are evaluated.

Broader pattern of non-compliance emerges

This oversight is not an isolated incident. A recent academic study reviewed thousands of federal AI products and found that more than 100 were out of compliance with federal or administrative regulations. These included tools used by various departments, such as the EPA’s system for prioritizing hazardous waste inspections. Researchers emphasize that in a democracy, the public must understand how high-stakes decisions are made to ensure that AI tools remain trustworthy and serve the public interest.

Agency cites accuracy concerns in delay

In response to inquiries, the EPA stated that it is premature to release information because the tools are not yet fully deployed. The agency explained that chemical assessment is technically complex and that accuracy is essential to ensure regulatory actions remain protective. While the EPA declined to answer many specific questions, it emphasized that it is still deliberating on how to best deploy these tools. Critics argue that waiting until deployment is too late to address potential flaws or biases in the underlying models.