Fascinating implications. Information submitted in court or formal hearings has long been considered immune from libel accusations - because such things are stated or submitted under oath, it is fair for an author or publisher to assume them to be true. That standard may now shift.
Of course, that does raise the potential pathway forward for lawyers, academics, and others who might rely on AI to augment their arguments made in such formal proceedings. These statements and submissions are made under the penalty of perjury. While it may seem harsh, the lesson perhaps should be taught. If some academic is going to submit AI generated data to a Senate panel and not fact-check it, they should face the penalties. "Sorry, AI made me do it," is not an excuse.
The line I have always drawn with AI is not regarding ethics. It's economics. You cannot responsibly compute a risk assessment for AI since its output is an unknown (if you could know it, you wouldn't need AI). These people who want to save a little time or money on the front end may find themselves facing time in prison or a substantial fine on the back end.
