AI RESEARCH
Don't Click That: Teaching Web Agents to Resist Deceptive Interfaces
arXiv CS.AI
•
ArXi:2605.09497v1 Announce Type: new Vision-language model (VLM) based web agents nstrate impressive autonomous GUI interaction but remain vulnerable to deceptive interface elements. Existing approaches either detect deception without task integration or document attacks without proposing defenses. We formalize deception-aware web agent defense and propose DUDE (Deceptive UI Detector & Evaluator), a two-stage framework combining hybrid-reward learning with asymmetric penalties and experience summarization to distill failure patterns into transferable guidance. We.