A.I. in Environmental Permitting
Senate Bill 1046 would create a Joint A.I. Environmental Permitting Program within the Environmental Management Commission and the Department of Environmental Quality to use artificial intelligence as a decision-support tool in reviewing and drafting environmental permits. The bill defines artificial intelligence broadly to include large language models, natural language processing, and retrieval-augmented generation systems, and it authorizes the Commission to adopt rules governing permissible uses, quality control, staff training, confidentiality, cybersecurity, and restrictions on training vendor-owned models with applicant or agency data unless specifically authorized.
The bill is explicit that AI may assist, but may not replace, human judgment: no permit application could be approved, denied, delayed, conditioned, or otherwise acted on solely because of AI output, and agency staff must independently review and modify AI-generated work product as needed. It also requires public transparency about the AI systems used, notice to applicants when AI is involved, annual reporting to legislative oversight bodies, and phased implementation beginning with post-construction stormwater permits, then air permits, then erosion and sedimentation control permits, before any broader expansion. The act would take effect July 1, 2026 and includes a $1 million nonrecurring General Fund appropriation for implementation.
The bill would add a new statutory section to Chapter 143B governing environmental permitting operations at DEQ and the Environmental Management Commission, while also affecting permitting processes under Chapters 113A, 130A, and 143. It would not change the substantive legal standards for issuing, denying, modifying, or revoking permits, but it would create a new framework for how permit applications and draft permits are reviewed and prepared, including rulemaking authority, reporting duties, and implementation safeguards. It also provides funding and staffing flexibility for temporary AI expertise to support rollout of the program.
Based on the bill text, the overall tone is strongly supportive of using AI to improve government efficiency, reduce processing times, and increase consistency in permit review. The findings emphasize benefits to applicants, regulated entities, state employees, and the public, while preserving human oversight and legal compliance. No committee transcripts or votes were provided, so there is no recorded legislative debate or vote history to indicate broader support or opposition beyond the bill’s stated policy rationale.
The main points of potential contention are the use of AI in a regulatory setting, data privacy and cybersecurity, and the risk of relying too heavily on automated systems in permit decisions. The bill addresses these concerns by requiring independent human review, limiting AI to a support role, allowing the Commission to set confidentiality and security rules, and restricting training of third-party models on agency or applicant data without authorization. Another likely issue is implementation scope and timing, since the program expands in phases and depends on the Commission’s determination that rules, quality controls, and training are in place before each phase begins.