OpenAI disclosed disrupting two AI-powered influence operations linked to Russia and Iran, revealing efforts to manipulate global media through fake journalist personas, fabricated think tanks, and AI-generated content. The operations targeted Latin America and U.S.-Iran relations, with the Russian campaign rated as the highest-risk influence effort OpenAI has disrupted.
How the Russian Operation Worked: Fake Think Tanks and AI-Generated Propaganda
The Russian operation, named Dark Clark by OpenAI, centered on a fabricated research organization called the Social Research Center (SRC), which appeared to have real employees in Latin America. The group used AI tools like ChatGPT to create internal reports, generate content, and orchestrate campaigns targeting countries in the region. One notable effort involved a Peruvian outlet reporting that an event did take place at a Peruvian school. OpenAI identified nearly 100 articles published under fake bylines, many of which were distributed through small to medium-sized outlets focused on international affairs. The operation also produced fake documents, audio scripts, and misleading material, including claims about Ukraine’s honorary consul in Panama.
For example, the SRC was linked to a fake persona named “Mia Clark,” used to present the organization as legitimate. OpenAI rated the operation as Category 5, the highest on its scale, citing its complexity and the potential for widespread influence.
The Russian campaign also involved fictitious bylines, whose articles appeared in about a dozen publications worldwide. According to OpenAI’s findings, the operation was similar to a 2020 Russian effort posing as an independent news outlet linked to Yevgeny Prigozhin’s Internet Research Agency, which influenced the 2016 U.S. election.
In addition to its focus on Latin America, the Russian operation targeted Ecuador with two fakes in March, including a fake audio clip attributed to Ukraine’s consul. The campaign also used AI to draft internal reports, which described efforts to undermine Ukraine’s reputation and influence local politics. OpenAI noted that the Russian operation’s internal reports included explicit instructions to reference Stepan Bandera, a figure viewed as an independence hero by some Ukrainians but associated with fascism by others. A Peruvian outlet, Extra, reported that one such event did take place at a Peruvian school.
The Iranian Campaign: AI-Powered Fabrication and Global Media Infiltration
Iranian operatives used AI to create seven fake journalist personas, which they deployed to pitch articles about the U.S.-Iran war to online publications. These articles, numbering nearly 100, were published or syndicated in about a dozen outlets. The operation also generated social media comments in English and Persian, though OpenAI found little evidence of significant engagement. The Iranian campaign, rated Category 4, involved AI tools to refine articles and create editorial pitches, but the bulk of the work was done by human actors.
The technology helped create content that looked and felt and read authentic to the point where it was being published by legitimate outlets,
according to Darren Linvill. The Iranian campaign also involved internal reports that exaggerated their social media impact, using deceptive metrics to appear more effective than they were.
The accounts accumulated nearly 80,000 followers during the first half of 2026, though OpenAI cautioned this number should not be interpreted as 80,000 Americans being persuaded by the operation. The Iranian operatives created false narratives about the U.S.-Iran conflict using these personas and saturated social media with comments on geopolitical issues. Both operations made heavy use of AI to draft internal reports.
Broader Implications: AI’s Role in Modern Influence Campaigns
The disruptions highlight the growing role of AI in foreign influence operations, where tools like ChatGPT are used to automate tasks such as content creation, account management, and social media coordination. OpenAI’s findings align with broader concerns about how AI can be weaponized to amplify disinformation, as seen in recent campaigns by Iran, China, and private Israeli firms. Many of the posts generated by these campaigns failed to gain significant traction, with engagement often limited to single or double digits.
OpenAI also disrupted two Chinese-linked operations in June 2026, one focused on a data centre-focused campaign and the other aimed at criticizing U.S. tariffs. These instances heighten concerns for tech platforms and markets as fabricated identities and AI-generated content grow more accessible to create and spread. OpenAI said it blocked both groups from using its models, emphasizing the need for ongoing vigilance against AI-enabled disinformation.
OpenAI has provided information about both campaigns to relevant authorities, but the absence of definitive attribution—OpenAI noted it couldn't identify the origin of the operation or confirm if those involved were linked to the Iranian government—leaves uncertainty about the future of such efforts.
What’s Next? OpenAI’s Continued Monitoring and Industry Response
OpenAI has pledged to continue monitoring AI-enabled influence operations, emphasizing that its reports aim to further research and disruption
of similar schemes. Meanwhile, the U.S. government has taken steps to address AI’s dual-use potential, with the Trump administration having largely favored voluntary safeguards over sweeping federal regulations. However, critics argue that self-regulation alone may not be sufficient, as evidenced by the recent Reuters/Ipsos survey showing public skepticism about the Trump administration’s and Congress’s handling of AI risks.
OpenAI stated that reporting on AI-driven influence operations like those exposed Thursday will aid in preventing similar schemes in the future, and noted that these reports have occasionally deterred such activities.