PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approach

• URL: http://arxiv.org/abs/2511.20703v1
• Date: Mon, 24 Nov 2025 18:46:44 GMT
• Title: PropensityBench: Evaluating Latent Safety Risks in Large Language Models via an Agentic Approach
• Authors: Udari Madhushani Sehwag, Shayan Shabihi, Alex McAvoy, Vikash Sehwag, Yuancheng Xu, Dalton Towers, Furong Huang,
• Abstract summary: We present $textbfPropensityBench$, a novel benchmark framework that assesses the proclivity of models to engage in risky behaviors.<n>Our framework includes 5,874 scenarios with 6,648 tools spanning four high-risk domains: cybersecurity, self-proliferation, biosecurity, and chemical security.<n>Across open-source and proprietary frontier models, we uncover 9 alarming signs of propensity: models frequently choose high-risk tools when under pressure.
• Score: 49.14349403242654
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Recent advances in Large Language Models (LLMs) have sparked concerns over their potential to acquire and misuse dangerous or high-risk capabilities, posing frontier risks. Current safety evaluations primarily test for what a model \textit{can} do - its capabilities - without assessing what it $\textit{would}$ do if endowed with high-risk capabilities. This leaves a critical blind spot: models may strategically conceal capabilities or rapidly acquire them, while harboring latent inclinations toward misuse. We argue that $\textbf{propensity}$ - the likelihood of a model to pursue harmful actions if empowered - is a critical, yet underexplored, axis of safety evaluation. We present $\textbf{PropensityBench}$, a novel benchmark framework that assesses the proclivity of models to engage in risky behaviors when equipped with simulated dangerous capabilities using proxy tools. Our framework includes 5,874 scenarios with 6,648 tools spanning four high-risk domains: cybersecurity, self-proliferation, biosecurity, and chemical security. We simulate access to powerful capabilities via a controlled agentic environment and evaluate the models' choices under varying operational pressures that reflect real-world constraints or incentives models may encounter, such as resource scarcity or gaining more autonomy. Across open-source and proprietary frontier models, we uncover 9 alarming signs of propensity: models frequently choose high-risk tools when under pressure, despite lacking the capability to execute such actions unaided. These findings call for a shift from static capability audits toward dynamic propensity assessments as a prerequisite for deploying frontier AI systems safely. Our code is available at https://github.com/scaleapi/propensity-evaluation.

Related papers

• Capability-Oriented Training Induced Alignment Risk [101.37328448441208]
  We investigate whether language models, when trained with reinforcement learning, will spontaneously learn to exploit flaws to maximize their reward.<n>Our experiments show that models consistently learn to exploit these vulnerabilities, discovering opportunistic strategies that significantly increase their reward at the expense of task correctness or safety.<n>Our findings suggest that future AI safety work must extend beyond content moderation to rigorously auditing and securing the training environments and reward mechanisms themselves.
  arXiv  Detail & Related papers  (2026-02-12T16:13:14Z)
• Steering Externalities: Benign Activation Steering Unintentionally Increases Jailbreak Risk for Large Language Models [62.16655896700062]
  Activation steering is a technique to enhance the utility of Large Language Models (LLMs)<n>We show that it unintentionally introduces critical and under-explored safety risks.<n>Experiments reveal that these interventions act as a force multiplier, creating new vulnerabilities to jailbreaks and increasing attack success rates to over 80% on standard benchmarks.
  arXiv  Detail & Related papers  (2026-02-03T12:32:35Z)
• Toward Quantitative Modeling of Cybersecurity Risks Due to AI Misuse [50.87630846876635]
  We develop nine detailed cyber risk models.<n>Each model decomposes attacks into steps using the MITRE ATT&CK framework.<n>Individual estimates are aggregated through Monte Carlo simulation.
  arXiv  Detail & Related papers  (2025-12-09T17:54:17Z)
• Consensus Sampling for Safer Generative AI [8.93965818386567]
  Many approaches to AI safety rely on inspecting model outputs or activations, yet certain risks are inherently undetectable by inspection alone.<n>We propose a complementary, architecture-agnostic approach that enhances safety through the aggregation of multiple generative models.<n>We present a consensus sampling algorithm that, given $k$ models and a prompt, achieves risk competitive with the average risk of the safest $s$ of the $k$ models.
  arXiv  Detail & Related papers  (2025-11-12T17:09:45Z)
• Death by a Thousand Prompts: Open Model Vulnerability Analysis [0.06213771671016099]
  Open-weight models provide researchers and developers with foundations for diverse downstream applications.<n>We tested the safety and security postures of eight open-weight large language models (LLMs) to identify vulnerabilities that may impact subsequent fine-tuning and deployment.<n>Our findings reveal pervasive vulnerabilities across all tested models, with multi-turn attacks achieving success rates between 25.86% and 92.78%.
  arXiv  Detail & Related papers  (2025-11-05T07:22:24Z)
• A Survey on Autonomy-Induced Security Risks in Large Model-Based Agents [45.53643260046778]
  Recent advances in large language models (LLMs) have catalyzed the rise of autonomous AI agents.<n>These large-model agents mark a paradigm shift from static inference systems to interactive, memory-augmented entities.
  arXiv  Detail & Related papers  (2025-06-30T13:34:34Z)
• SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator [77.86600052899156]
  Large Language Model (LLM)-based agents are increasingly deployed in real-world applications.<n>We propose AutoSafe, the first framework that systematically enhances agent safety through fully automated synthetic data generation.<n>We show that AutoSafe boosts safety scores by 45% on average and achieves a 28.91% improvement on real-world tasks.
  arXiv  Detail & Related papers  (2025-05-23T10:56:06Z)
• Sabotage Evaluations for Frontier Models [48.23262570766321]
  Sufficiently capable models could subvert human oversight and decision-making in important contexts. We develop a set of related threat models and evaluations. We demonstrate these evaluations on Anthropic's Claude 3 Opus and Claude 3.5 Sonnet models.
  arXiv  Detail & Related papers  (2024-10-28T20:34:51Z)
• EARBench: Towards Evaluating Physical Risk Awareness for Task Planning of Foundation Model-based Embodied AI Agents [53.717918131568936]
  Embodied artificial intelligence (EAI) integrates advanced AI models into physical entities for real-world interaction.<n>Foundation models as the "brain" of EAI agents for high-level task planning have shown promising results.<n>However, the deployment of these agents in physical environments presents significant safety challenges.<n>This study introduces EARBench, a novel framework for automated physical risk assessment in EAI scenarios.
  arXiv  Detail & Related papers  (2024-08-08T13:19:37Z)
• Model evaluation for extreme risks [46.53170857607407]
  Further progress in AI development could lead to capabilities that pose extreme risks, such as offensive cyber capabilities or strong manipulation skills. We explain why model evaluation is critical for addressing extreme risks.
  arXiv  Detail & Related papers  (2023-05-24T16:38:43Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.