News

New Concerns Over AI Chatbot Falsehoods

Well, that didn’t take long: the excitement over generative AI tools like ChatGPT has already been replaced by growing concerns over their dangers.  The Center for Countering Digital Hate found it could easily bypass the guardrails in Google Bard to generate unlabeled misinformation in 78 of 100 tested cases.  And, moving from hypothetical to actual, an Australian politician is threatening to sue OpenAI after ChatGPT falsely reported that he was a convicted criminal.