SafePred: A Predictive Guardrail for Computer-Using Agents via World Models

Add code
Feb 02, 2026

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: