SafeEvalAgent: Toward Agentic and Self-Evolving Safety Evaluation of LLMs

• URL: http://arxiv.org/abs/2509.26100v1
• Date: Tue, 30 Sep 2025 11:20:41 GMT
• Title: SafeEvalAgent: Toward Agentic and Self-Evolving Safety Evaluation of LLMs
• Authors: Yixu Wang, Xin Wang, Yang Yao, Xinyuan Li, Yan Teng, Xingjun Ma, Yingchun Wang,
• Abstract summary: This paper introduces a new paradigm of agentic safety evaluation, reframing evaluation as a continuous and self-evolving process.<n>We propose a novel multi-agent framework SafeEvalAgent, which autonomously ingests unstructured policy documents to generate and perpetually evolve a comprehensive safety benchmark.<n>Our experiments demonstrate the effectiveness of SafeEvalAgent, showing a consistent decline in model safety as the evaluation hardens.
• Score: 37.82193156438782
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: The rapid integration of Large Language Models (LLMs) into high-stakes domains necessitates reliable safety and compliance evaluation. However, existing static benchmarks are ill-equipped to address the dynamic nature of AI risks and evolving regulations, creating a critical safety gap. This paper introduces a new paradigm of agentic safety evaluation, reframing evaluation as a continuous and self-evolving process rather than a one-time audit. We then propose a novel multi-agent framework SafeEvalAgent, which autonomously ingests unstructured policy documents to generate and perpetually evolve a comprehensive safety benchmark. SafeEvalAgent leverages a synergistic pipeline of specialized agents and incorporates a Self-evolving Evaluation loop, where the system learns from evaluation results to craft progressively more sophisticated and targeted test cases. Our experiments demonstrate the effectiveness of SafeEvalAgent, showing a consistent decline in model safety as the evaluation hardens. For instance, GPT-5's safety rate on the EU AI Act drops from 72.50% to 36.36% over successive iterations. These findings reveal the limitations of static assessments and highlight our framework's ability to uncover deep vulnerabilities missed by traditional methods, underscoring the urgent need for dynamic evaluation ecosystems to ensure the safe and responsible deployment of advanced AI.

Related papers

• Beyond single-channel agentic benchmarking [0.0]
  This paper argues that evaluating AI agents in isolation systematically mischaracterizes their operational safety when deployed within human-in-the-loop environments.<n>Even imperfect AI systems can nonetheless provide substantial safety utility by functioning as redundant audit layers against well-documented sources of human failure.
  arXiv  Detail & Related papers  (2026-02-05T08:22:02Z)
• SecureCAI: Injection-Resilient LLM Assistants for Cybersecurity Operations [0.0]
  This paper introduces SecureCAI, a novel defense framework extending Constitutional AI principles with security-aware guardrails.<n>SecureCAI reduces attack success rates by 94.7% compared to baseline models.
  arXiv  Detail & Related papers  (2026-01-12T18:59:45Z)
• DeepKnown-Guard: A Proprietary Model-Based Safety Response Framework for AI Agents [12.054307827384415]
  Large Language Models (LLMs) have become increasingly prominent, severely constraining their trustworthy deployment in critical domains.<n>This paper proposes a novel safety response framework designed to safeguard LLMs at both the input and output levels.
  arXiv  Detail & Related papers  (2025-11-05T03:04:35Z)
• OpenAgentSafety: A Comprehensive Framework for Evaluating Real-World AI Agent Safety [58.201189860217724]
  We introduce OpenAgentSafety, a comprehensive framework for evaluating agent behavior across eight critical risk categories.<n>Unlike prior work, our framework evaluates agents that interact with real tools, including web browsers, code execution environments, file systems, bash shells, and messaging platforms.<n>It combines rule-based analysis with LLM-as-judge assessments to detect both overt and subtle unsafe behaviors.
  arXiv  Detail & Related papers  (2025-07-08T16:18:54Z)
• IS-Bench: Evaluating Interactive Safety of VLM-Driven Embodied Agents in Daily Household Tasks [30.535665641990114]
  We present IS-Bench, the first multi-modal benchmark designed for interactive safety.<n>It features 161 challenging scenarios with 388 unique safety risks instantiated in a high-fidelity simulator.<n>It facilitates a novel process-oriented evaluation that verifies whether risk mitigation actions are performed before/after specific risk-prone steps.
  arXiv  Detail & Related papers  (2025-06-19T15:34:46Z)
• AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions [64.85086226439954]
  We present SAFE, a benchmark for assessing the safety of embodied VLM agents on hazardous instructions.<n> SAFE comprises three components: SAFE-THOR, SAFE-VERSE, and SAFE-DIAGNOSE.<n>We uncover systematic failures in translating hazard recognition into safe planning and execution.
  arXiv  Detail & Related papers  (2025-06-17T16:37:35Z)
• 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)
• strideSEA: A STRIDE-centric Security Evaluation Approach [1.996354642790599]
  strideSEA integrates STRIDE as the central classification scheme into the security activities of threat modeling, attack scenario analysis, risk analysis, and countermeasure recommendation.<n>The application of strideSEA is demonstrated in a real-world online immunization system case study.
  arXiv  Detail & Related papers  (2025-03-24T18:00:17Z)
• Agent-SafetyBench: Evaluating the Safety of LLM Agents [72.92604341646691]
  We introduce Agent-SafetyBench, a benchmark designed to evaluate the safety of large language models (LLMs)<n>Agent-SafetyBench encompasses 349 interaction environments and 2,000 test cases, evaluating 8 categories of safety risks and covering 10 common failure modes frequently encountered in unsafe interactions.<n>Our evaluation of 16 popular LLM agents reveals a concerning result: none of the agents achieves a safety score above 60%.
  arXiv  Detail & Related papers  (2024-12-19T02:35:15Z)
• 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)

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.