Jessica A. Baweja

This report examines the application of artificial intelligence (AI) technologies for insider threat mitigation (ITM) programs in nuclear security facilities. Insider threat detection presents unique challenges due to the subtle and adaptive nature of these threats, the complex signatures involved, and the scarcity of available data for analysis. Traditional human-centered approaches, while essential, face limitations in processing large amounts of data continuously and detecting subtle patterns across multiple systems. AI technologies can potentially address these limitations by providing 24/7 monitoring capabilities, identifying complex patterns that might escape human observation, and offering consistent application of security criteria. However, the deployment of AI in nuclear security contexts introduces significant new risks, including workflow disruption, expanded attack surfaces, potential for misuse, and ethical concerns regarding privacy, fairness, transparency, safety, and security. The high-consequence nature of nuclear security decisions demands careful consideration of these risks and systematic approaches to their mitigation.

Artificial intelligence (AI) has the potential to transform nuclear security operations, offering opportunities to enhance effectiveness while simultaneously introducing new challenges. As AI technologies rapidly evolve, agencies across the United States Government (USG) are researching, implementing, and evaluating various AI models and systems. Given the broad capabilities and applications of these technologies, it is essential for each agency to identify and articulate those areas where it can make meaningful contributions aligned with its mission and expertise. To address this need for strategic focus, in late Fiscal Year 2025 (FY2025), the Office of International Nuclear Security (INS) established an AI Task Force (AITF) to gather input from subject matter experts (SMEs) regarding the most appropriate role INS could serve in researching, evaluating, or implementing AI for nuclear security. The AITF engaged 15 experts from national laboratories with backgrounds in cyber security, physical security, transport security, insider threat mitigation, nuclear engineering, human-systems engineering, and AI/ML development. This white paper summarizes the insights gathered from these SMEs and presents a potential roadmap for INS engagement with AI technologies. The recommendations outlined here are intended to inform INS leadership as they make strategic decisions about resource allocation and program direction in this rapidly evolving technological domain.

High-stakes analytical environments require analysts to evaluate evidence and generate conclusions to inform critical decisions often under conditions of uncertainty. Probabilistic decision-making based on incomplete or inaccurate information can reduce productivity, compromise national interests, and endanger public safety. Researchers are developing expert systems built on foundation models (FMs) to support analysts’ decision-making processes by enabling human-artificial intelligence (AI) teaming, in part through the quantification and expression of uncertainty information. As FMs continue to mature, it is imperative to correspondingly consider analysts’ needs for appropriately interpreting and using uncertainty information. However, prior research indicates that it remains unclear how analysts engage with FM-generated uncertainty information and the extent to which these interactions influence trust in, and reliance on, expert systems. We plan to review the state of the science and conduct an exploratory, qualitative study to (a) understand how properly communicated uncertainty can foster calibrated trust and appropriate reliance and (b) identify approaches for effectively conveying FM-generated uncertainty information during analytical workflows. We will administer semi-structured interviews with analysts from a specific high-stakes analytical environment to collect their current experiences with job-related uncertainty and their impressions when viewing FM-generated uncertainty information. During the interview protocol, participants will be presented with several different FM outputs and invited to discuss their thoughts and beliefs about the uncertainty information displayed. Participants may provide insights into how trust and reliance may be influenced by uncertainty. The results of this study will help us to better understand how analysts currently interpret and use uncertainty information. Our findings may inform human factors recommendations for effectively conveying uncertainty information to foster calibrated trust in, and appropriate reliance on, expert systems. Interaction designers and FM developers can use this knowledge to enhance human-AI teaming and ensure the responsible deployment of FM-based expert systems in analytical workflows.

This 2-year work plan provides a roadmap for activities the IWG and other stakeholders will undertake between 2025 and 2027. Members of the IWG Steering Committee (SC) are responsible to (1) engage the broader insider threat mitigation community of practice to inform and promote work plan activities, (2) track progress towards goals, and (3) provide updates to the IWG at regular intervals. This work plan is a living document and will be updated periodically to assess progress towards goals and will be revised as needed to address emergent and exigent areas of concern identified by the IWG.

In this study, we investigated strategies to address trust issues arising from errors in large language models (LLMs). The study examined the impact of confidence scores, system capability explanations, and user feedback on trust restoration post-error. 68 participants viewed the responses of an LLM to 20 general trivia questions, with an error introduced on the third trial. Each participant was presented with one mitigation strategy. Participants rated their overall trust in the model and the reliability of the answer. Results showed an immediate drop in trust after the error; however, there were no differences across the three strategies in trust recovery. All conditions had a logarithmic trend in trust recovery following error. Differences in overall trust were predicted by perceived reliability of the answer, suggesting that participants were evaluating results critically and using that to inform their trust in the model. Qualitative data supported this finding; participants expressed lasting distrust despite the LLM’s later accuracy. Results showcase the need to prioritize accuracy in LLM deployment, because early errors may irrevocably damage user trust calibration and later adoption.

The rapid expansion of generative artificial intelligence (GenAI) has generated excitement regarding its potential benefits and concern over its ethical implications. Governments, corporations, and standards organizations have described ethical principles to direct GenAI's development and use; however, practical guidance for implementing these principles is limited. Addressing this gap is critical, especially considering the array of risks associated with GenAI, such as legal liabilities, privacy concerns, security threats, and potential misuse. Robust policies and procedures are critical to support responsible deployment of GenAI. This report examines Pacific Northwest National Laboratory (PNNL)’s approach to promoting responsible GenAI use. Proposed initiatives include developing policies based on ethical principles, creating a governance process to review projects relative to those principles, and implementing onboarding processes for training staff. The governance framework described in this report adapts the structure and principles of Institutional Review Boards (IRBs), traditionally used in human subjects research, for GenAI ethical review, providing oversight. Ethical principles guiding responsible GenAI usage include transparency and accountability, privacy, fairness, safety, security, and validity and reliability. To operationalize these principles, we propose forming a GenAI Assurance Council (GAC) that mirrors the IRB's structure. The GAC will evaluate GenAI projects across privacy, accountability, transparency, safety, security, fairness, and validity dimensions. Complementing policy and governance is AI literacy training to support staff understanding of GenAI's ethical implications. An initial training effort for AI Incubator Chat—a GenAI tool deployed at PNNL—showed promising results, underscoring the importance of clear guidelines and user accountability. Collaborative efforts and the dissemination of best practices are also discussed. The proposed GAC model and AI literacy training provide a blueprint for establishing ethical GenAI use and governance, offering practical tools to bridge the gap between ethical principles and real-world applications. The responsible integration of GenAI at PNNL entails a multifaceted approach involving policy development, ethical governance, and AI literacy training. The positive initial feedback and collaborative opportunities position PNNL to lead by example in GenAI's responsible use, reflecting a proactive stance in addressing the ethical, legal, and societal challenges associated with this emerging technology. PNNL's systematic and ethical approach to GenAI offers a model for other institutions to emulate, promoting safe and responsible technological advancements in the AI domain.

Domestic violent extremism has been on the rise in recent years, fueled by growing political polarization, the COVID-19 pandemic, and the spread of misinformation online. This concerning trend raises an important question: has this increase in extremist violence manifested in more incidents involving radioactive or nuclear (RN) materials and a corresponding security concern? To explore this question, we analyzed existing data on events involving chemical, biological, radiological, or nuclear materials over time. Specifically, we examined the rate of RN-related incidents, the targets of these events, and the ideologies motivating the perpetrators by using the Violent Non-State Actor Chemical, Biological, Radiological, and Nuclear Event Database from the National Consortium for the Study of Terrorism and Responses to Terrorism. Our analysis aimed to determine whether the broader rise of violent extremism has translated to an elevated risk of attacks involving RN materials. Results of the analysis revealed that there was no clear increase in events involving RN material in the dataset. However, given the rarity of these events, this conclusion does not indicate that such a trend might not appear in coming years if violent extremism continues to rise. To address the potential risk presented by violent extremism, in addition to quantifying the frequency of RN events, we also reviewed traits that could make organizations more susceptible to insider threats from radicalized employees. Focusing on the literature around counterproductive workplace behavior, we identified organizational characteristics that might increase the risk of malicious acts committed by radicalized insiders. These risk factors include abusive leadership, situational constraints, and organizational injustice. Finally, using these organizational risk characteristics, we discuss interventions that organizations with RN material can take to reduce their risk of attacks stemming from violent extremism while still protecting individual rights to privacy and civil liberties. Overall, this article provides an overview of the intersections between rising domestic extremism, insider threats, and risks posed by RN materials. We present current data on trends and patterns in this space as well as practical guidance for organizational resilience.

Corrigendum: Opportunities for human factors in machine learning

A novel set of system-state and control-action penalty functions are introduced as an alternative to traditional performance index contingency ranking. The novel system state penalty metrics are formulated based on piecewise linear functions of the system voltage and branch flow, guided by Weber’s Law of human cognition. Novel continuous and discrete control action metrics are also developed to measure the inherent cost and risk associated with every action taken by human power system operator to resolve violations on a pre-contingent basis. These new metrics are combined with traditional human factors indices for measuring human-machine trust and cognitive workload to create a systematic framework for measuring and evaluating operator trust and reliance on artificial intelligence (AI) algorithms for control room use. An existing AI-based contingency analysis recommender tool using a semi-supervised action algorithm is selected for a series of experiments with operations engineering staff using the IEEE 118 Bus System. The penalty metrics presented are demonstrated for both steady-state contingency analysis and transient stability studies, with the operations participants able to reduce the total system penalty in 85% of scenarios through remedial actions. A human-machine team was able to achieve equal or lower continuous control action penalty scores than the participant without availability of the recommender in 57% of experiment scenarios and lower continuous control action penalty scores than the AI tool alone in 83% of scenarios.