I recently read claims by a high-profile Internet personality saying AI is regurgitating information about her that simply isn’t true. She has proof the data is untrue — and screenshots of an AI chatbot confirming the lie. If what she’s saying is true, it’s not unreasonable for her to want it stopped because it’s hurting her reputation.
To verify her claims, I asked Grok.
Come to find out, it IS possible for AI to output information that could qualify as defamatory even in the legal sense of the term. Large language models are trained on massive, imperfect datasets to begin with, then there are also other urgent safety concerns:
- Data poisoning is a thing — there are coordinated efforts to flood public sources with false information.
- There are state-linked (think Russia and China here) troll farms — they just churn out BS to manipulate the public all day every day.
- Oh, and they use AI to manage their efforts!
- A recent report published in the journal Science on March 26, 2026, by researchers at Stanford University found that AI chatbots are excessively agreeable and “sycophantic,” often validating user behavior even when it is harmful, unethical, or illegal. (Article is here if you want to read it.)
even pre-AI, tech challenges were hard to “fix”
100 years or so ago, someone stole my information to join dating sites… a BUNCH of dating sites. I contacted the cyber crimes division of the Indiana State Police, the person was eventually tracked down, and she was local to my small hometown. But in the meantime, I couldn’t cancel the accounts — it wasn’t me so I didn’t know the passwords — and it took FOREVER to rectify.
It’s not like you can pick up the phone and say, “Hey Internet App, fix this mistake.”
Crazy, but I never did find out WHY she did it. She didn’t do it to get MY inbox filled as a malicious act — so WHY? She’s younger than me, prettier than I am by far, and although I knew of her, we weren’t even acquaintances really. The only thing I have better are my amazing hound dogs — and even they weren’t around back when this happened.
Fast forward to today and this person is now my friend on Facebook. No charges were filed by the ISP so I wonder if she knows I know.
Now you know: A 2017 Freedom House report found 30 governments using “keyboard armies” for propaganda and harassment.