External brief: a more efficient regulator, still accountable

Unofficial support material for the public and other outside stakeholders. This is not an NRC press release, not an NRC policy, and not a use of the NRC seal. It repeats what the agency has already published, and it says when a figure is not an artificial-intelligence result.

Audience. Members of the public, license applicants, and other outside stakeholders who want a plain account of how the NRC is using artificial intelligence, automation, and emerging technology to work more efficiently.

The public commitment

The NRC's enterprisewide AI vision is to responsibly leverage artificial intelligence so staff can efficiently make risk-informed decisions, drive regulatory innovation, and protect public health, safety, and the environment. [C02]

That efficiency claim is about assistance. The agency's public inventory says AI is not a substitute for professional judgment or legal interpretation. People remain accountable for the decision. [C10]

The fiscal year 2026 AI Strategic Plan, published on September 29, 2025, says new technology is meant to support efficient, risk-informed, and reliable regulatory outcomes. [C03]

The strategy also says AI tools can speed access to historical and new information, streamline workflows, and create efficiencies across the enterprise, and that broader use waits on a phased evaluation. [C08]

What the public can already see

Anyone can read the use-case inventory. Two cases are deployed.

SimplifAI, use case NRC-0001, is deployed for text retrieval and generation. The NRC says it supports more consistent licensing and oversight work and more efficient use of institutional knowledge. [C11]

Microsoft 365 Copilot and Copilot Chat, use case NRC-0002, are deployed to automate repetitive tasks and streamline data analysis so staff time can move to higher-value work. That is the automation the agency has described, not a claim that reviews run themselves. [C12]

Search is part of the same story. The NRC says AI-powered cognitive search produced ADAMS Content Search and significantly streamlined search for staff. The public version, ADAMS Public Search, launched in January 2025. [C16]

A licensing page says staff are using Copilot Chat to research precedents and compare documents, and that an internal generative AI tool is being piloted on a new enrichment facility application. No public percentage is attached to that pilot. [C15]

Emerging tools, said at their real stage

USAi, use case NRC-0003, is still in pre-deployment. It is a planned secure doorway to several language models. It is not a finished product the public should treat as already speeding reviews. [C13]

A separate test of ChatGPT Enterprise and Gemini for Government, use case NRC-0004, is also pre-deployment. The NRC says the test has helped staff draft and summarize non-sensitive material and has shaped the rules for any wider use. [C14]

The Artificial Intelligence Compliance Plan, issued in September 2025 under OMB Memorandum M-25-21, says this access is supposed to stay inside enterprise licenses, FedRAMP-authorized environments, and NRC-approved services. [C04]

For the next 12 months, the agency has said it will broaden generative AI training for staff and contractors and set rules of behavior. [C21]

Oversight of that work sits with a Chief AI Officer, an AI Governance Board, and an AI Steering Committee. [C06]

A September 2025 briefing says 2025 delivered SimplifAI as the first internal AI tool, and it describes 2026 expansion toward acceptance reviews, licensing, inspections, and mission support. The later item is a plan. [C24]

The RIC 2026 session on modernizing mission work describes automated workflows and intelligent document processing as productivity tools. That description is not a published savings percentage. [C22]

Preparing to review AI used by the industry

Efficiency is also about being ready to review other people's AI without a new rule for every case. An October 2024 assessment looked at NRC regulations and more than 500 Regulatory Guides and found the framework generally sufficient, with some places that may need clearer guidance. [C17]

NUREG-2261, published in May 2023, is the earlier readiness plan for reviewing AI in regulated activities effectively and efficiently. It is not an internal time-savings report. [C18]

The Future Focused Research program, launched in FY 2020, includes public examples on inspection planning, cybersecurity states, and reading regulatory documents with natural language processing. [C19]

What outsiders should not be told

Revised milestone schedules, aligned to the 12-month and 18-month periods cited in Executive Order 14300, took effect on May 23, 2025. Those clocks are process and policy reforms. They are not an AI result. [C23]

If a briefing, a poster, or a conversation blends those schedules into the AI story, it is wrong.

Where to read the record

The claim ids in this brief resolve in `claims/claim-list.md`. Start with the inventory at the internally focused AI page, the enterprisewide strategy, and the licensing efficiencies page, which is also the page that separates the AI pilot from the schedule reforms.