4. LLMs are going to be used both by con artists (“I’m going to fool this judge with nonexistent precedent!”) and con victims (“The AI found precedent for me, and I believe it!”) — and those two are sometimes going to be almost indistinguishable. Many stories like this are going to live in a murky “Is it deceptive if they let themselves be fooled?” zone. Various professions need ethical guidelines, fast.