I feel like this actually highlights an undocumented feature of AI that institutions are taking advantage of: the ability to go "The AI fucked up. It happens. What're you gonna do?" I don't see anyone being held responsible and I don't see any plan to prevent this in the future.