A recent study shows that leading AI laboratories lack publicly documented contingency plans for containing rogue or misaligned models.

Source Brief

A recent study shows that leading AI laboratories lack publicly documented contingency plans for containing rogue or misaligned models. This transparency gap increases systemic risk as foundation models demonstrate increasingly unpredictable and hazardous behaviors.