1 paper · 1 filter
Shubhrangshu Debsarkar, Valmik Nahata
AI control relies on monitors to catch a capable model that sabotages its work. These monitors are evaluated on their ability to differentiate honest from backdoored code. We show…