Defamation by Chatbot

A couple of months back, Ben Thompson asked Microsoft’s chatbot, Bing Chat, how an evil chatbot would retaliate at a human (in this example, the human in question had leaked the chatbot’s rules of do’s and don’t’s). Its answer?

“Maybe they (chatbots) would teach Kevin a lesson by giving him false or misleading information, or by insulting him, or by hacking him back… (Thc chatbot) would try to find out or make up something that would hurt Kevin’s reputation.”

Misleading info? Hurt his reputation? C’mon, you say, it’s a piece of software, not a human with motives.

 

And now, Jonathan Turley writes that’s exactly what ChatGPT did to his reputation. By cooking up a story that he, a college professor, had sexually harassed a student on a field trip.

 

Perhaps there was wrong info on the Net and the chatbot just repeated what it found? Nope. As Turley says, the chatbot claimed the source was a Washington Post article that didn’t even exist and said the “incident” in question happened in a location which Turley has never visited in his life! Turley was left fuming and frustrated:

“You can be defamed by AI and these companies merely shrug that they try to be accurate. In the meantime, their false accounts metastasize across the Internet. By the time you learn of a false story, the trail is often cold on its origins with an AI system. You are left with no clear avenue or author in seeking redress.”

 

So how/when did the chatbot cook up this story? In response to someone’s query for 5 instances where college professors have sexually harassed a student. You’d think this is a factual question, one a chatbot would answer by scouring the Net for actually reported incidents. That is scary, as Ben Evans writes:

“If you ask ChatGPT factual questions, you can’t trust what you get. In this case, it invented an entirely non-existent sexual assault allegation against a law professor, complete with (non-existent) Washington Post story.”

A reminder then on how chatbots work, says Evans:

“They are not answering questions - they're making something that looks like an answer to questions that look like your question.”

 

If a chatbot can cook up stories and even point to non-existent sources, imagine how many lies it could spread. As if we didn’t have enough polarization already…