Russian fake think-tank with real employees exposed in ‘most complex’ influence operation attempt, OpenAI says
A separate Iran-origin campaign used ChatGPT to edit and pitch nearly 100 articles that were published by real media outlets – under fake journalist bylines.
The OpenAI logo is seen in this illustration created on Jun 11, 2026. (File photo: Reuters/Dado Ruvic/Illustration)
New: You can now listen to articles.
This audio is generated by an AI tool.
Read a summary of this article on FAST.
Get bite-sized news via a new
cards interface. Give it a try.
Click here to return to FAST
Tap here to return to FAST
FAST
SINGAPORE: OpenAI on Thursday (Oct 8) said it banned clusters of accounts traced to Russia and Iran, for using its ChatGPT AI assistant to support covert influence operations. The Russian campaign in particular was the first to have likely led to public comment by politicians in several countries, the US-based generative AI firm added.
In a report, OpenAI said the accounts had combined the use of AI with more conventional deceptive techniques to run what it dubbed “false front” operations.
The Russian operation allegedly involved the use of a fake think-tank targeting Latin American countries and staffed by unwitting local employees; while the Iranian operation was believed to have used fake journalist personas to pitch articles to genuine media outlets around the world.
In both cases, OpenAI said the campaigns were used to “launder geopolitical, conflict-related messaging into their target audiences”, with ChatGPT used to create both internal and public-facing content.
Guess Word
Crack the word, one row at a time
Buzzword
Create words using the given letters
Mini Sudoku
Tiny puzzle, mighty brain teaser
Mini Crossword
Small grid, big challenge
Word Search
Spot as many words as you can
OpenAI did not disclose the number of accounts involved or identify the ultimate operators behind them.
HIGHEST-IMPACT INFLUENCE OPERATION
OpenAI’s Oct 8 report identified a key arm of the Russian-origin operation as running a front research organisation called the Social Research Center. References to this entity were found in internal reports generated by the campaign’s operators using ChatGPT.
The campaign appeared aimed at damaging perceptions of Ukraine in Latin America, but also extended to domestic interference in several regional countries.
The reports indicated that the operators had tricked locals in Latin America into working for the SRC and creating original content for it, by interviewing regional experts and writing reports that were published on the SRC website.
“The available evidence suggests that these on-the-ground employees had no idea that they were working for a Russian operation, and conducted research tasks in good faith,” said OpenAI.
It added that the operation was “the most complex attempt to run a front identity” that it had disrupted in the past two-and-a-half years.
OpenAI rated the Russian campaign a five — the second-highest level — on a scale developed by researchers to measure the impact of influence operations. This was the first Category 5 operation OpenAI had disrupted since it began publishing such reports, with evidence the campaign had prompted public comment by politicians in a number of countries.
The campaign operators’ internal reports also referenced other wide-ranging influence activity, including alleged attempts to spread fake “leaked” audio recordings in Bolivia.
OpenAI could not corroborate all of these claims, and it noted that the operators had also tried to take credit for activity they were not responsible for.
“In other reports, the operators asked our models to help identify incidents in the target countries that they could take credit for, even if they had not been involved,” OpenAI said.
FAKE JOURNALISTS
The Iranian operation meanwhile involved the use of ChatGPT to refine articles about the US-Iran conflict and to subsequently draft email pitches to real media outlets, said OpenAI, suggesting the activity appeared consistent with that of a hired commercial actor.
OpenAI identified nearly 100 articles that were published or syndicated to about a dozen outlets between July 2025 and October 2026. These were written under seven fictitious bylines, using the personas of American or Western journalists covering foreign affairs and geopolitics.
Some personas had accompanying social media accounts to create the impression of legitimacy, but in at least one case, platform transparency settings showed the accounts were based in Iran.
Some articles were also promoted by the social media accounts of the media outlets, “giving the Iranian operators a potential way to access audiences that they would not otherwise have reached”, added OpenAI, which rated this aspect of the campaign a Category 4 on the influence operations scale.
A less successful strand of the campaign involved using ChatGPT to generate social media comments on the US-Iran war in both English and Persian. OpenAI said the comments it found typically drew likes, views and replies in only single or double digits, with many coming from the operation’s own accounts.
“In fact, the operators’ internal reports used a deceptive calculation to measure their social media impact, possibly in an attempt to inflate the operators’ apparent effectiveness,” it added.
OpenAI has periodically reported on influence operations it has detected in the past. In June 2026, it said it disrupted two Chinese-linked operations, one which ran a data centre-focused campaign and the other aimed at criticising US tariffs.
Source: CNA/jo
Sign up for our newsletters
Get the CNA app
Stay updated with notifications for breaking news and our best stories
Get WhatsApp alerts
Join our channel for the top reads for the day on your preferred chat app
Get bite-sized news via a new
cards interface. Give it a try.
Click here to return to FAST
Tap here to return to FAST
FAST
