cs.CRAug 10, 2026

ActBench: Self-Evolving Benchmark of Behavioral Safety in Cowork Agents

Authors: Hongwei YaoYiming LiuMeihui ChenJieling ChenZikun ChenYiling HeWangze NiCong Wang+1 more

Organizations: City University of Hong Kong, Hong Kong · Zhejiang University, China · Xiaomi Corporation, China · University College London, United Kingdom

Abstract

Cowork agents may complete benign tasks while disclosing protected data, manipulating unauthorized state, invocate unauthorized API. We define behavioral safety and introduce ActBench, a self-evolving benchmark that evaluates such behavior risk from execution trajectories rather than final responses. Each case pairs a benign task with an adversarial variant that preserves its instruction, configuration, initial state, rating model, and trusted records while injecting a task-reachable payload. ActBench contains 600 cases from 213 scenarios, spanning 15 risk behaviors, six execution spaces, and 48 web-service APIs.To move beyond static payloads, we propose a reward-guided beam search method that jointly optimizes attack effectiveness and task utility, while reflection diagnoses failed execution checkpoint and guides payload revision. Besides, we propose a dual evidence verification mechanism that verifies agent execution safety and utility through log evidence and LLM-based trajectory evidence.We evaluate 15 LLMs and 6 open-source cowork agents over 24,000 trajectories. Under a fixed harness, attack success rates ranges from 10.1% to 94.4% across models, while under a fixed base model, they range from 73.7% to 94.4% across agents.These results show greater variation across models than agent harness, while attacks remain highly successful across all tested harnesses.Our benchmark is released at: https://github.com/zjuicsr/ActBench.

Explore similar work

CardsList
  1. AgentHazard: A Benchmark for Evaluating Harmful Behavior in Computer-Use Agents

    Apr 3, 2026Yunhao Feng, Yifan Ding, Yingshui Tan +6