Blab AI
NRC Cuts Nuclear Licensing Time by Half Using AI, Says Deputy AI Officer
In a breakthrough that could reshape nuclear regulation, the U.S. Nuclear Regulatory Commission announced that artificial‑intelligence tools have slashed the time required to issue a nuclear facility license from four years to nine months. The reduction came after a pilot licensing case that began in 2024 and wrapped up in early 2026, according to the agency’s Chief Data Officer and Deputy Chief AI Officer, Basia Sall.
Sall told attendees of the ATARC AI for Impact summit in Virginia that the agency’s progress was partly driven by a 2025 executive order from President Donald Trump, which set a federal deadline of 18 months for licensing reviews. Building on that mandate, the NRC’s own AI initiatives pushed the timeline even further, enabling the pilot case to finish in nine months.
The NRC’s AI strategy has been layered on recent regulatory changes and federal guidance. The agency has used AI to draft regulatory documents—ensuring that precedent decisions are considered—and to analyze public data sets supplied by industry partners. By allowing external developers to curate NRC public data for their own applications, the agency receives what Sall described as “a much better application than we have in the past.”
In addition to its internal tools, the NRC has tapped the General Services Administration’s OneGov program, which launched in April 2025. OneGov offers agencies discounted access to private‑sector technologies, and the NRC has tested Anthropic’s Claude, Azure OpenAI, and Google Gemini for limited use cases involving public data. According to GSA, more than 120 orders for AI offerings have been placed through OneGov, giving access to roughly 3.4 million federal employees.
The NRC is also beginning to use the GSA‑sponsored USAi platform, launched in August 2025 as a testing ground for AI tools. A GSA official noted that over 25 agencies already use USAi, with 16 more expected to join before the end of 2026.
Internally, the NRC has developed a tool called SimplifAI, built on Azure OpenAI. The agency upgraded to a 2.0 version after the initial model was deprecated. According to the NRC’s most recent AI use‑case inventory, SimplifAI’s text‑retrieval and generation capabilities improve efficiency and consistency in licensing, oversight, and other regulatory activities. Some NRC employees have trained SimplifAI to help draft speeches.
Sall emphasized that the agency’s approach is to maintain a “menu of tools” rather than rely on a single solution. She added that the NRC is exploring additional AI use cases beyond licensing, including document drafting and internal communications.
The NRC’s progress reflects a broader trend of federal agencies adopting AI to accelerate regulatory processes. While the agency’s licensing time has already been cut in half, officials believe further efficiencies are possible as AI tools mature and additional platforms become available.
The NRC’s experience also illustrates the practical impact of the GSA’s OneGov and USAi initiatives, which aim to reduce costs and speed innovation across federal agencies. As the NRC continues to refine its AI strategy, it will likely serve as a model for other regulatory bodies seeking to balance speed, safety, and compliance.
In summary, the NRC has successfully used AI to reduce a nuclear licensing case from four years to nine months, building on a 2025 executive order and leveraging GSA programs and internal tools. The agency plans to expand its AI toolset and explore new applications, while maintaining rigorous safety and regulatory standards.
Sall told attendees of the ATARC AI for Impact summit in Virginia that the agency’s progress was partly driven by a 2025 executive order from President Donald Trump, which set a federal deadline of 18 months for licensing reviews. Building on that mandate, the NRC’s own AI initiatives pushed the timeline even further, enabling the pilot case to finish in nine months.
The NRC’s AI strategy has been layered on recent regulatory changes and federal guidance. The agency has used AI to draft regulatory documents—ensuring that precedent decisions are considered—and to analyze public data sets supplied by industry partners. By allowing external developers to curate NRC public data for their own applications, the agency receives what Sall described as “a much better application than we have in the past.”
In addition to its internal tools, the NRC has tapped the General Services Administration’s OneGov program, which launched in April 2025. OneGov offers agencies discounted access to private‑sector technologies, and the NRC has tested Anthropic’s Claude, Azure OpenAI, and Google Gemini for limited use cases involving public data. According to GSA, more than 120 orders for AI offerings have been placed through OneGov, giving access to roughly 3.4 million federal employees.
The NRC is also beginning to use the GSA‑sponsored USAi platform, launched in August 2025 as a testing ground for AI tools. A GSA official noted that over 25 agencies already use USAi, with 16 more expected to join before the end of 2026.
Internally, the NRC has developed a tool called SimplifAI, built on Azure OpenAI. The agency upgraded to a 2.0 version after the initial model was deprecated. According to the NRC’s most recent AI use‑case inventory, SimplifAI’s text‑retrieval and generation capabilities improve efficiency and consistency in licensing, oversight, and other regulatory activities. Some NRC employees have trained SimplifAI to help draft speeches.
Sall emphasized that the agency’s approach is to maintain a “menu of tools” rather than rely on a single solution. She added that the NRC is exploring additional AI use cases beyond licensing, including document drafting and internal communications.
The NRC’s progress reflects a broader trend of federal agencies adopting AI to accelerate regulatory processes. While the agency’s licensing time has already been cut in half, officials believe further efficiencies are possible as AI tools mature and additional platforms become available.
The NRC’s experience also illustrates the practical impact of the GSA’s OneGov and USAi initiatives, which aim to reduce costs and speed innovation across federal agencies. As the NRC continues to refine its AI strategy, it will likely serve as a model for other regulatory bodies seeking to balance speed, safety, and compliance.
In summary, the NRC has successfully used AI to reduce a nuclear licensing case from four years to nine months, building on a 2025 executive order and leveraging GSA programs and internal tools. The agency plans to expand its AI toolset and explore new applications, while maintaining rigorous safety and regulatory standards.
