算法裂变：为AI服务而削弱的核监管

The Undermining of NuclearRegulation in Service of AI Dr. Sofia Guerra, FREng and Dr. Heidy KhlaafNovember 2025 Table of Contents Introduction and Overview3 Key Developments in Civilian Nuclear Regulation4 Relegating Nuclear Thresholds for the AI Arms Race Critiquing the Critiques: The LNT Model and ALARAPrinciple6 The Use of Generative AI and Risks to Nuclear Safety,Security, and Nonproliferation12 Nuclear Licensing: Processes for Permitting NuclearFacilities21 Oversimplification and Risks of Using Generative AI forNuclear Licensing24 Risks to Nuclear Safeguarding and Proliferation New Nuclear Technologies and Overstated Promises toPower AI30 Introduction and Overview As the AI industry’s insatiable energy demands collide with power grid infrastructure limits—with anexpected 160 percent increase in data center power demand due to generative AI by 20301—AIcompanies have set their sights on nuclear energy as a source from which they can extract a colossalfive tofifty gigawatts of additional power by 2028.2AI labs have thus begun mounting pressure toaccelerate the deployment of nuclear energy sources, with major nuclear initiatives underway in anattempt to meet this recent surge in demand. These AI demands are currently infeasible, as nucleardevelopment timelines—often ten to twenty years—are out of step with the pace of AI deployment,with large conventional nuclear reactors only capable of producing up to one gigawatt of power (i.e.,1GW[e]) per unit. This discrepancy between the AI industry’s energy demands and the lack of technicalfeasibility to construct nuclear plants at the pace of these stringent (and often contrived) timescaleshas created a chasm that is ultimately leading to a slew of efforts to fast-track nuclear timelines thatraise serious safety and oversight concerns. Although the nuclear sector has the opportunity toexpedite global decarbonization efforts, the monopolization of nuclear energy to explicitly power AIraises serious concerns about whether the risks associated with nuclear facilities and unsubstantiated,fast-tracked initiatives can be justified if they are not to the benefit of civil energy consumption, and ifthey further entrench power asymmetries that may lead to nuclear destabilization and proliferation. This report taxonomizes and assesses these nuclear “fast-tracking” initiatives. We examine theirfeasibility and their impact on nuclear safety, security, and safeguards3—and more largely society’spotential exposure to radiation levels—across three primary categories: 1.Policy initiatives seeking to lower regulatory practices including long-established nuclear-safetyand risk-analysis approaches, safety culture, acceptable risks, and thresholds in order to reducetimescales for the construction of civil and defense nuclear facilities 2.The use of generative AI to expedite regulatory processes such as nuclear licensing andcommissioning for both civil and defense nuclear facilities 3.The promotion of advanced and new nuclear technologies that are contingent on novel orunmaterialized approaches and infeasible timescales First, policy initiatives are being introduced to lower regulatory oversight in order to expedite theconstruction of civil nuclear facilities. AI labs’ assertions regarding the urgency of immediate energyneeded for AI has put unprecedented pressure on regulators to reconsider well-establishednuclear-safety and risk-analysis approaches, such as the linear no-threshold (LNT) model for radiationexposure and the “as low as reasonably achievable” (ALARA) risk principle with a lack ofwell-researched and tried-and-tested alternatives to replace these standards.4,5These initiatives aresimultaneously accompanied by the reduced independence of nuclear regulatory bodies,6justified byalleged national-security imperatives tied to a purported AI arms race. However, such a politicization of nuclear regulation will ultimately lead to the skewing of cost-benefit analysis that may result inincreased risk tolerances to society’s potential exposure to radiation levels. That is, the use of nuclearenergy to power the development of generative AI further increases the risk of the public’s exposure toionizing radiation without a clear or substantiated benefit to justify it. Furthermore, the unprecedentedtrend of AI labs directly investing in the very nuclear providers they intend to utilize to exclusivelypower their data centers may lead to conflicts of interest that compromise nuclear licensee readinessand expectations in terms of organizational capabilities and safety culture. Second, AI-based proposals (and even deployment) have been put forward by AI labs,7nuclearproviders,8and licensees9to use large language models (LLMs) to generate nuclear regulatory andlicensing documents in hopes of expediting nuclear licensing and commissioning processes. Suchefforts purportedly claim that generative-AI will “analys[e] historic nuclear licensing data [that] allowslicensing engineers