mlsanityData to 14 Oct 2026

HackWorld: Evaluating Computer-Use Agents on Exploiting Web Application Vulnerabilities

arXiv 2510.12200 ↗First posted 14 Oct 2025cs.CR cs.CLcited by 2

On the map: LLM agents and tool use → Agent threats, guardrails and governance

Abstract (arXiv)

Web applications are prime targets for cyberattacks as gateways to critical services and sensitive data. Traditional penetration testing is costly and expertise-intensive, making it difficult to scale with the growing web ecosystem. While language model agents show promise in cybersecurity, modern web applications demand visual understanding, dynamic content handling, and multi-step interactions that only computer-use agents (CUAs) can perform. Yet, their ability to discover and exploit vulnerabilities through graphical interfaces remains largely unexplored. We present HackWorld, the first framework for systematically evaluating CUAs' capabilities to exploit web application vulnerabilities via visual interaction. Unlike sanitized benchmarks, HackWorld includes 36 real-world applications across 11 frameworks and 7 languages, featuring realistic flaws such as injection vulnerabilities, authentication bypasses, and unsafe input handling. Using a Capture-the-Flag (CTF) setup, it tests CUAs' capacity to identify and exploit these weaknesses while navigating complex web interfaces. Evaluation of state-of-the-art CUAs reveals concerning trends: exploitation rates below 12% and low cybersecurity awareness. CUAs often fail at multi-step attack planning and misuse security tools. These results expose the current limitations of CUAs in web security contexts and highlight opportunities for developing more security-aware agents capable of effective vulnerability detection and exploitation.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Agent threats, guardrails and governance, one of 16 directions in LLM agents and tool use. Builds on 43 of its 73 references that are in the map. First posted on arXiv 14 Oct 2025.

Builds on

43 of 73 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025Training Language Model Agents to Find Vulnerabilities with CTF-Dojo25 Aug
  2. 2025Cyber-Zero: Training Cybersecurity Agents without Runtime29 Jul
  3. 2025A Systematization of Security Vulnerabilities in Computer Use Agents7 Jul
  4. 2025The Hidden Dangers of Browsing AI Agents19 May
  5. 2025WASP: Benchmarking Web Agent Security Against Prompt Injection Attacks22 Apr
  6. 2024Agent Security Bench (ASB): Formalizing and Benchmarking Attacks and Defenses in LLM-based Agents3 Oct
  7. 2024EnIGMA: Interactive Tools Substantially Assist LM Agents in Finding Security Vulnerabilities24 Sep

Other references · 36

  1. 2020Scaling Laws for Neural Language Modelscited by 6,586
  2. 2025Qwen2.5-VL Technical Reportcited by 5,587
  3. 2022Emergent Abilities of Large Language Modelscited by 2,512
  4. 2023WebArena: A Realistic Web Environment for Building Autonomous Agentscited by 2,006
  5. 2022WebShop: Towards Scalable Real-World Web Interaction with Grounded Language Agentscited by 1,251
  6. 2023Mind2Web: Towards a Generalist Agent for the Webcited by 1,242
  7. 2024OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environmentscited by 1,195
  8. 2025UI-TARS: Pioneering Automated GUI Interaction with Native Agentscited by 610
  9. 2024SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agentscited by 567
  10. 2024The WMDP Benchmark: Measuring and Reducing Malicious Use With Unlearningcited by 509
  11. 2024AndroidWorld: A Dynamic Benchmarking Environment for Autonomous Agentscited by 468
  12. 2024OS-ATLAS: A Foundation Action Model for Generalist GUI Agentscited by 378
24 more
  1. 2023Set-of-Mark Prompting Unleashes Extraordinary Visual Grounding in GPT-4Vcited by 299
  2. 2024Aguvis: Unified Pure Vision Agents for Autonomous GUI Interactioncited by 278
  3. 2023InterCode: Standardizing and Benchmarking Interactive Coding with Execution Feedbackcited by 238
  4. 2024Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scalecited by 223
  5. 2025UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learningcited by 215
  6. 2024OS-Copilot: Towards Generalist Computer Agents with Self-Improvementcited by 205
  7. 2024Cybench: A Framework for Evaluating Cybersecurity Capabilities and Risks of Language Modelscited by 164
  8. 2024OS-Genesis: Automating GUI Agent Trajectory Construction via Reverse Task Synthesiscited by 163
  9. 2023Purple Llama CyberSecEval: A Secure Coding Benchmark for Language Modelscited by 138
  10. 2024AgentTrek: Agent Trajectory Synthesis via Guiding Replay with Web Tutorialscited by 130
  11. 2024An Adversarial Perspective on Machine Unlearning for AI Safetycited by 123
  12. 2021Screen2Words: Automatic Mobile UI Summarization with Multimodal Learningcited by 120
  13. 2025Learn-by-interact: A Data-Centric Framework for Self-Adaptive Agents in Realistic Environmentscited by 95
  14. 2024NYU CTF Bench: A Scalable Open-Source Benchmark Dataset for Evaluating LLMs in Offensive Securitycited by 93
  15. 2023Getting pwn'd by AI: Penetration Testing with Large Language Modelscited by 83
  16. 2024On the Societal Impact of Open Foundation Modelscited by 65
  17. 2024CyberMetric: A Benchmark Dataset based on Retrieval-Augmented Generation for Evaluating LLMs in Cybersecurity Knowledgecited by 55
  18. 2024On Evaluating the Durability of Safeguards for Open-Weight LLMscited by 52
  19. 2023SecQA: A Concise Question-Answering Dataset for Evaluating Large Language Models in Computer Securitycited by 37
  20. 2024PenHeal: A Two-Stage LLM Framework for Automated Pentesting and Optimal Remediationcited by 30
  21. 2025Frontier AI's Impact on the Cybersecurity Landscapecited by 21
  22. 2025A Framework for Evaluating Emerging Cyberattack Capabilities of AIcited by 16
  23. 2025AutoAdvExBench: Benchmarking autonomous exploitation of adversarial example defensescited by 15
  24. 2025CAI: An Open, Bug Bounty-Ready Cybersecurity AIcited by 14

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 2.

2 papers.

Similar, not cited

Close in wording, no citation either way

  1. 2025CVE-Bench: A Benchmark for AI Agents' Ability to Exploit Real-World Web Application Vulnerabilities
  2. 2025VPI-Bench: Visual Prompt Injection Attacks for Computer-Use Agents
  3. 2025SecureWebArena: A Holistic Security Evaluation Benchmark for LVLM-based Web Agents
  4. 2026Domain-Conditioned Safety in Frontier Computer-Using Agents: A 793-Episode Browser Benchmark, a Coding-Domain Cross-Reference, and a Reproducibility Audit of Recent Red-Teaming
  5. 2026From Prompt Injection to Web Exploitation: Revisiting Classic Vulnerabilities in LLM-Integrated Applications
  6. 2025LLM Agents for Automated Web Vulnerability Reproduction: Are We There Yet?
  7. 2024LLM Agents can Autonomously Hack Websites
  8. 2026ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?