Usually you spot the fake by the language. Here you spot it by the library.
Fact
The facts: on 25 August 2026 OpenAI reported that it had banned a cluster of ChatGPT accounts which, in its assessment, very likely originated in Russia. The operators prompted in Russian and received social media comments posted on Substack, Telegram, X, Facebook and LinkedIn, most of them in English. The content promoted the site of a self-described expert community called the International Burke Institute, registered in February 2025 with a street address in Israel. One operator also produced German-language posts for a Telegram channel; another generated logos for a dozen channels covering Germany, the USA, France, Poland and Türkiye. By OpenAI's assessment the impact was limited: low view counts and low subscriber numbers.
How you build authority out of nothing
The site listed names of the rank of Francis Fukuyama and Noam Chomsky. The articles underneath were real, but they belonged to other people.
One, on the China-Pakistan Economic Corridor, appears copied from a Cambridge University Press original and credited to a different academic. Another, on migration governance, was taken from the Migration Policy Institute and signed with the name of an Australian professor of food chemistry. The photograph on her page is the same one that sits on a lecture listing of hers from 2017.
A stolen paper does not convince the reader. It fills the shelf so the shelf looks full.
This is stagecraft, not argument. Nobody comes in to read a paper about Pakistan. They come in, see many headlines and respected names, and accept that the place is serious. From there they read the rankings, which is what the whole thing was built for.
What stays with me
Not the scale. The reach was small and OpenAI says so plainly.
What stays is the division of labour. The model did the part that wants volume and patience: comments, replies under other people's posts, logos for a dozen channels. The content carrying the message came from elsewhere and was human work.
For a long time we said broken language was the cheapest fraud detector there is. There is no broken language here, but there is something more telling: a source you can check in a minute. The original of the paper is online, with the real name under it.
Official primary sources
OpenAI: Disrupting a new covert influence campaign from Russia (official report)The visual is generated code-art, no third-party images.