RedTeamCUA: Realistic Adversarial Testing of Computer-Use Agents in Hybrid Web-OS Environments
Zeyi Liao , Jaylen Jones , Linxi Jiang , Yuting Ning , Eric Fosler‑Lussier , Yu Su , Zhiqiang Lin , Huan Sun
- 🏛 Institutions
- OSU
- 📅 Date
- May 28, 2025
- 📑 Publisher
- ICLR 2026 (Oral)
- 💻 Env
- Desktop Web
- 🔑 Keywords
TLDR
RedTeamCUA introduces a hybrid OS-and-web sandbox for realistic adversarial testing of computer-use agents under indirect prompt injection. Its RTC-Bench benchmark contains 864 hybrid attack scenarios and shows that current frontier agents still exhibit substantial attack success rates in both initialized and end-to-end settings.
Related papers (24)
- StepJack: Benchmarking Computer-Use Agent Safety Against Multi-Step Indirect Prompt InjectionAugust 6, 2026 · arXiv
- LoginTrap: Uncovering Task-Agnostic Phishing-Style Indirect Prompt Injection Attacks against LLM-based Web AgentsAugust 5, 2026 · arXiv
- The Blind Spot of Agent Safety: How Benign User Instructions Expose Critical Vulnerabilities in Computer-Use AgentsApril 12, 2026 · arXiv
- WebSP-Eval: Evaluating Web Agents on Website Security and Privacy TasksApril 7, 2026 · arXiv
- HackWorld: Evaluating Computer-Use Agents on Exploiting Web Application VulnerabilitiesOctober 14, 2025 · ICLR 2026 (Poster)
- VPI-Bench: Visual Prompt Injection Attacks for Computer-Use AgentsJune 3, 2025 · ICLR 2026 (Poster)
- WASP: Benchmarking Web Agent Security Against Prompt Injection AttacksApril 22, 2025 · NeurIPS 2025 (Poster)
- Are Android GUI Agents Robust Against Runtime Anomalies? AnTrap: Evaluating Agents in Dynamic Adversarial EnvironmentsAugust 25, 2026 · arXiv
- OSReward: Instituting Standardized Evaluation for Cross-Platform Computer-Use Reward ModelsJuly 30, 2026 · arXiv
- VenusBench-GD: A Comprehensive Multi-Platform GUI Benchmark for Diverse Grounding TasksDecember 18, 2025 · arXiv
- OS-Oracle: A Comprehensive Framework for Cross-Platform GUI Critic ModelsDecember 18, 2025 · arXiv
- RiOSWorld: Benchmarking the Risk of Multimodal Computer-Use AgentsMay 31, 2025 · NeurIPS 2025 (Poster)
- On the Robustness of GUI Grounding Models Against Image AttacksApril 7, 2025 · arXiv
- sudo rm -rf agentic_securityMarch 26, 2025 · ACL 2025 Industry Track
- In-Context Defense in Computer Agents: An Empirical StudyMarch 12, 2025 · arXiv
- WorldGUI: An Interactive Benchmark for Desktop GUI Automation from Any Starting PointFebruary 12, 2025 · arXiv
- GUI Action Narrator: Where and When Did That Action Take Place?June 19, 2024 · arXiv
- GUI-World: A Video Benchmark and Dataset for Multimodal GUI-oriented UnderstandingJune 16, 2024 · ICLR 2025 (Poster)
- OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer EnvironmentsApril 11, 2024 · NeurIPS 2024 Datasets and Benchmarks Track
- OmniACT: A Dataset and Benchmark for Enabling Multimodal Generalist Autonomous Agents for Desktop and WebFebruary 29, 2024 · ECCV 2024 (Poster)
- SeeClick: Harnessing GUI Grounding for Advanced Visual GUI AgentsJanuary 17, 2024 · ACL 2024
- VRL-Bench: Benchmarking agents on computer control tasks under finite trial budgetsSeptember 11, 2026 · arXiv
- JarvisGUI: Towards Cross-Device GUI Agents with Dynamic Task CompositionSeptember 9, 2026 · arXiv
- FinCUABuild: Can Agents Build Reliable Benchmarks for Dynamic Financial Computer Use?September 7, 2026 · arXiv