Which would bring you nowhere. If they didn't change this at some point, I remember at the time everyone was staring to use ChatGPT that OpenAI wrote in their terms that the user is responsible for the model's output. If they can do this, I expect other model providers doing this as well.

Is there precedent where that clause was enforced?

No idea.