For CISOs, the sharper question is not whether frontier AI creates fresh danger, but whether it exposes long-standing identity, patching, and recovery gaps faster than teams can close them.