Threats from using AI to spread misinformation, deepfakes and child sexual abuse materials will only get worse without meaningful regulation. Some frameworks have been proposed, but they do not include ways to hold AI developers accountable for the misuse of their systems.