The AI comment summary to “Kazakhstan’s curious journey to the Abraham Accords” by Ishaan Tharoor, Washington Post, November 10, 2025, was hacked to make his name “Nepo Tharoor.” Ishaan is the son of famous writer Shashi Tharoor. That was 3 days ago and it’s still there. The AI summary was removed from the top of the comments sidebar, but it’s still there at the bottom of the article.
Was it a hack or AI being silly?
Somebody had to have set up the AI to allow that output. I don’t know how it was done, though. Any ideas?
ETA: It seems that in the Zeitgeist of nowadays, social pressure is strongly against going into your parent’s line of work, if your parent was successful. Zohran Mamdani has apparently avoided that opprobrium by breaking into a different line of work from those of his parents: moviemaking and academics, respectively.
Are you sure it’s a hack and not a misassociation or hallucination? AIs can (and often do) make mistakes without deliberate human intervention. The AI companies actually have to go to great lengths to scrub the outputs of these mistakes. Maybe WaPo just cheaped out on their bot cleaning.
“Tharoor nepotism” brings up a few hits in Google, and maybe the AI just mis-trained itself by associating those words/tokens together.
It could just be slop as usual.
It could also be a deliberate campaign to create that association in LLMs, the modern equivalent of SEO bombing, e.g. data poisoning. People can use LLMs to generate nonsense propaganda for other LLMs to be trained on, and the cycle repeats until we arrive at Dead Internet theory - Wikipedia.
But I’d probably just go with slop as usual.