Yesterday I've mentioned the current challenge with LLMs/GPT to ensure accurate answers.
Today I've found this: ChatGPT gave a terrible answer to the question "What role did Brian Hood have in the Securency bribery saga?", defaming a completely innocent person.
In another instance, criminal defense attorney Jonathan Turley was falsely reported by ChatGPT of sexually harassing a student and cited a Washington Post article that was never published.
I wonder if that may cause the service to be closed and the use of LLMs be restricted by law to certain uses (say, it could be restricted from mentioning anyone).