There are countless stories of fraudsters using AI to social engineer or directly defraud companies. Feels obvious that AI can be used to harden systems/defenses. I just don't see that much of it.