Frontier AI labs still won’t say how they’d contain a rogue model

A recent study indicates that major AI laboratories lack publicly documented strategies for managing or containing models that exhibit unexpected, harmful behaviors. This transparency gap highlights concerns regarding industry preparedness as increasingly autonomous systems approach capabilities that could pose significant safety risks.