MaskClaw: Edge-Side Personalized Privacy Arbitration for GUI Agents with Behavior-Driven Skill Evolution

Researchers introduced MaskClaw, an edge-side privacy arbitrator for GUI agents.

It uses local visual evidence and policy memory to decide whether to Allow, Mask, or Ask before screenshots leave a trusted environment.

User corrections are turned into reusable privacy skills, helping the system improve over time.

A new benchmark, P-GUI-Evo, shows current limits in pattern matching, cloud reasoning, and routing baselines.