AI monitoring can make companies cheat smarter, not stop.

Regulators using AI to catch rule-breakers might just make the rule-breakers better at hiding their violations. If the regulator doesn't invest enough in AI, or if human reviewers get overwhelmed, companies can keep finding new ways to break rules without getting caught.

This paper shows that simply adding AI to enforcement doesn't automatically mean better compliance. If the AI isn't good enough, or if the human part of the system can't keep up, companies might just learn to cheat in more sophisticated ways. This means regulators need to think carefully about how much they invest in AI and how they manage the human review process, or they could end up with more complex violations and no better results.