Security1 publisherNot yet confirmed elsewhere2 min readPublished
Russian and Iranian AI-made personas got propaganda into real news outlets before OpenAI shut them down
OpenAI shut down Russian and Iranian influence operations whose AI-built fake journalists and think tanks got stories into real news outlets. For editors who take outside pitches, a credible byline or institute name proved nothing about whether a real person stood behind it.
The Watch · Security desk
What happened
- The Russian network, which OpenAI calls Dark Clark, ran a supposed Latin American think tank, the Social Research Center, fronted by an AI-made leader named Mia Clark.
- OpenAI says the center's Latin American employees did not know they were working for a Russian group.
- During Bolivia's May protests, a fake audio clip claimed the government would cut water to La Paz under a state of emergency, and the government issued a denial.
- OpenAI scored the Russian campaign 5 and the Iranian one 4 on a six-point impact scale where few campaigns rise above 1 or 2.
Compiled by The WatchSomething wrong?How this is made
Why it matters
- constraint A check that a think tank employs real people producing original work would have passed the SRC, since OpenAI says its co-opted staff wrote most of the content.
- decision Editors who take contributed pieces on international affairs now have to decide how much identity checking a contributor gets before a byline runs, because seven invented reporters got past the checks those outlets used.
- contradiction CyberScoop calls the placements mainstream, but OpenAI describes the Iranian outlets as small to medium, and those stories often drew few comments or social reactions.
The two networks built their fake identities differently [3]. The Iranian operation invented the reporters outright. "We identified almost 100 articles published or syndicated under the operation's bylines across roughly a dozen online outlets around the world," OpenAI said [16]. Spread across at least seven personas, that comes to no more than about 14 articles per fake byline [20]. OpenAI said the activity resembled "a commercial actor running a for-hire influence campaign," and did not name an actor or group [18].
The Russian operation invented a leader and put real people to work under her. OpenAI called the Social Research Center "the most complex attempt to run a front identity that we've disrupted over the past two and a half years" [7]. Its evidence that the Russian operators controlled the center came from their own ChatGPT sessions, in which they discussed wage scales and decisions to hire and fire staff [6]. A publisher vetting the institute from outside would only have seen the people working there and what they produced.
The center's output mostly went after Ukraine's standing in Latin America, and it took up divisive political issues in Argentina and Bolivia [5]. In March, Lupa Media reported on a TikTok video falsely claiming Ecuador's government was using shell companies to recruit citizens to fight for Ukraine against Russia [10]. The operators also spread fake audio of the Ukrainian consul in Ecuador insulting Ecuadorians [11]. OpenAI points to fact-checks by large Latin American news organizations and denials by officials in the region that seem to be answering the network's material [19].
The examples CyberScoop gives of the Russian network reaching large outlets are those fact-checks and denials. CyberScoop reported that high impact ratings reflect exactly that response, the kind that tends to come when a narrative grows big enough that authorities must deny it [2].
OpenAI did not tie Dark Clark to a specific Russian government or intelligence agency [13]. Its operators asked the model about Politology, a successor to the Wagner Group once led by Yevgeny Prigozhin, "far more" than about any other Russian organization, OpenAI said [13].
What to watch
- Whether OpenAI or another investigator ties Dark Clark to a specific Russian body, given how often its operators asked about Politology.
- Whether the roughly dozen outlets that ran the Iranian bylines are named or retract the articles.
- Whether the for-hire operator behind the Iranian campaign shows up again under new journalist personas.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption
- Insufficient
- Hype gap+15
- Incentives
- Insufficient
- Confidence50
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
OpenAI rates campaigns on an impact scale of 1 to 6, with few rising above 1 or 2; it rated the Russian campaign a 5 and the Iranian campaign a 4.
ReportedSupportedSource: CyberScoop, citing OpenAI2 sources— create a free account to open themView cited source - [2]
The ratings indicate the campaigns were significantly more successful in breaking through to mainstream audiences, eliciting fact checks and official denials that tend to happen when a narrative becomes so big that authorities must deny it.
- [3]
OpenAI disclosed Thursday it shut down two influence operations from Russia and Iran that used ChatGPT and other AI tools to create fake journalist personas and covert think tanks that planted stories and narratives in mainstream news publications.
- [4]
The Russian cluster, which OpenAI calls 'Dark Clark', focused on Latin American politics and culture and is named for Mia Clark, a fake AI persona presented as the leader of the Social Research Center, a supposed Latin American think tank used to seed propaganda in news reporting.
- [5]
Dark Clark's content primarily focused on harming Ukraine's reputation in Latin America, though it also became involved in hot-button political issues in Argentina and Bolivia.
- [6]
Based on reports generated through ChatGPT in which operatives discussed wage scales for employees and hiring and firing decisions, OpenAI researchers said it is likely the Russian group was in control of the Social Research Center rather than being an employee or observer.
- [7]
OpenAI called it "the most complex attempt to run a front identity that we've disrupted over the past two and a half years."
- [8]
"The available evidence indicates that the SRC's employees in Latin America were not aware that they were working for a Russian group," OpenAI said.
- [9]
"Unlike another Russia-linked 'think tank' that we recently exposed, the SRC appears to have produced a majority of original content via its co-opted staff," OpenAI said.
- [10]
In March, Lupa Media reported on a TikTok video falsely claiming that the Ecuadorian government was using shell companies to recruit citizens to fight for Ukraine against Russia.
- [11]
The Russian actors generated and spread fake audio clips of the Ukrainian consul in Ecuador insulting Ecuadorian citizens.
- [12]
During anti-government protests in Bolivia in May, the group released a fake audio clip of an employee at the state water company claiming the government was about to declare a state of emergency and shut off water to La Paz, prompting an official denial from the Bolivian government.
- [13]
The report does not attribute the activity to any specific Russian government or intelligence agency; its operators prompted the model for information on Politology, a successor to the Wagner Group previously led by Yevgeny Prigozhin, "far more" than any other Russian organization.
- [14]
The Iranian campaign used AI tools to pitch written articles on the U.S.-Iran war; operators created at least seven fake journalist personas who placed dozens of stories online through legitimate media organizations.
- [15]
The Iranian operation's stories often failed to elicit many comments or social media reactions.
- [16]
"We identified almost 100 articles published or syndicated under the operation's bylines across roughly a dozen online outlets around the world," OpenAI said.
- [17]
"These were small to medium outlets, generally focused on international affairs, geopolitics, and events in the Middle East," OpenAI said of the Iranian placements.
- [18]
OpenAI said the Iranian activity resembled "a commercial actor running a for-hire influence campaign," but did not attribute it to a specific actor or group.
- [19]
OpenAI cites several fact-checks and official denials from major media organizations and government officials in Latin America that appear to respond to Dark Clark's propaganda.
- [20]
The Iranian operation averaged no more than about 14 articles per fake journalist persona.
Sources
1 independent publisher whose own reporting we read for this story.
- cyberscoop.comOpenAI says Iran, Russia used AI journalists, think tanks to influence Western media
2 articles · October 8, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.