Reward hacking will continue to be a problem. For sufficiently astonishing AI-created results, we must remember to ask if AI has not instead found it easier to elaborately fool us.