
ChatGPT and Bard Cheerfully Repeat Fake News Like It's Their Job
Ourtake
Two rounds of testing, same result: the bots keep repeating the rumors they were supposed to fact-check.
What happened
NewsGuard's repeat audit found OpenAI's ChatGPT-4 produced false claims for 98 of 100 news-related myths and Google's Bard for 80 of 100. The results were nearly identical to earlier tests, with no clear improvement reported.