I don’t see that as very different. Once it’s established that this is possible or even likely outcome of using AI at scale, we should treat this as a form of negligence and act accordingly.