Pacific Northwest National Laboratory and OpenAI partner to accelerate federal permitting

Pacific Northwest National Laboratory (PNNL), a U.S. Department of Energy national lab, has partnered with OpenAI to introduce DraftNEPABench, a novel benchmark designed to measure how effectively AI coding agents can accelerate the federal environmental permitting process. The collaboration focuses on the National Environmental Policy Act (NEPA), a foundational law requiring environmental reviews for major federal projects, which are often lengthy and complex. The benchmark provides a standardized way to test AI agents on tasks like drafting sections of Environmental Impact Statements (EIS), with initial results indicating these AI systems could reduce the time spent on certain drafting tasks by up to 15%. This represents a significant step in applying advanced AI to bureaucratic processes that can delay infrastructure, energy, and transportation projects for years.

The DraftNEPABench evaluates agents—AI systems that can take actions like writing and editing code or text—on their ability to understand regulatory requirements, analyze project data, and generate compliant draft documents. By creating a measurable standard, PNNL and OpenAI aim to drive development of AI tools that can handle the nuanced, document-intensive work of federal permitting. For government agencies and contractors, this could translate to faster project approvals for everything from clean energy grids to highway expansions. The partnership signals a growing focus on applying generative AI beyond consumer applications into complex, high-stakes regulatory and scientific domains, potentially setting a precedent for modernizing other government workflows with AI assistance.