arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2605.09684cs.CRcs.AI
收藏

MonitoringBench: 面向代理监控的半自动化红队测试

MonitoringBench: Semi-Automated Red-Teaming for Agent Monitoring

Independent(独立)

Monika Jotautaitė, Maria Angelica Martinez, Ollie Matthews, Tyler Tracy

AI总结:

本文提出半自动化红队方法,针对代理监控中的攻击检测问题,通过改进攻击生成和测试流程,构建了包含2644条攻击轨迹的MonitoringBench基准,评估监控能力与失败模式。

摘要:

我们介绍了一种红队方法,旨在暴露更难被捕捉的攻击,表明当前实践可能低估攻击效果并高估监控性能。我们识别出三个挑战:首先,攻击生成中的模式崩溃,通过新的攻击分类学解决;其次,构思与执行之间的差距,通过将攻击构建分解为策略生成、执行和事后轨迹优化来缓解;最后,手动获取成本高,通过半自动化红队流程解决。应用于BashArena,一个用于工具使用编码代理的AI控制环境,该流程生成MonitoringBench,包含2644条攻击轨迹以评估监控能力和失败模式。我们的流程产生更多样化和更强的攻击:Opus-4.5监控器在仅手动获取的Opus攻击中的捕捉率从94.9%下降到60.3%在我们最佳优化的攻击中,其他中等水平监控器下降更明显。针对三个开发监控器优化的攻击能泛化到十个测试监控器,捕捉率通常随监控能力增加。使用此基准,我们提供当前监控能力的快照,并发现前沿监控器常能检测可疑行为但易受说服或无法适当校准可疑性分数,表明改进路径可行。MonitoringBench为当前工具使用监控提供静态基准,并为代理和监控进步时刷新评估提供可重用的方法。

英文摘要:

We introduce a red-teaming methodology that exposes harder-to-catch attacks for coding-agent monitors, suggesting that current practices may under-elicit attacks and overstate monitor performance. We identify three challenges with current red-teaming. First, mode collapse in attack generation, which we reduce with a novel attack taxonomy for broader coverage. Second, a conceive-execute gap: frontier LLMs can propose strong attack ideas or execute them, but not all at once. We mitigate this by decomposing attack construction into strategy generation, execution, and post-hoc trajectory refinement. Third, manual elicitation is costly to scale, which we address with our semi-automated red-teaming pipeline. Applied to BashArena, an AI control setting for tool-using coding agents, this pipeline produces MonitoringBench, a benchmark of 2,644 attack trajectories for evaluating monitor capabilities and failure modes. Our pipeline produces more diverse and stronger attacks: Opus-4.5 monitor's catch rate falls from 94.9\% on elicited-only Opus attacks to 60.3\% on our best refined attacks, with larger drops for several mid-tier monitors. Attacks optimized against three development monitors generalize to ten held-out monitors, with catch rates generally increasing with monitor capability. Using this benchmark, we provide a snapshot of the current monitor capabilities and find that frontier monitors often detect suspicious actions but fall for persuasion or fail to calibrate suspiciousness scores appropriately, suggesting tractable paths for improvement. MonitoringBench provides both a static benchmark for current tool-use monitors and a reusable methodology for refreshing these evaluations as agents and monitors improve.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑