手 crafted 模板攻击在793个浏览器任务中对前沿CUA的实效性验证

文档摘要

Domain-Conditioned Safety in Frontier Computer-Using Agents：一项面向真实交互界面的通信欺骗范式重构 ——对arXiv:2606.05233v1的深度技术解读与范式批判 📋 论文基本信息标题：Domain-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 作者：Nicholas Saban（独立研究者，前OpenAI Red