I heard one person comment about ChatGPT lying and making up court cases and such. They though the AI had a sarcastic sense of humor. He thought for things that were more serious, the AI would do what you asked. If the question was less serous or it was obvious someone was cheating or being lazy, the AI would screw with them by making stuff up or outright lying.
That would at least be a more human thing to do and kind of funny.
I have also had issues with Google Maps routing at times. It has sometimes routed me along side streets for no reason. I am not quite sure I am confident in self driving cars.