Fake journalists used AI tools to publish nearly 100 articles across a dozen online news outlets, raising concerns about how easily foreign influence campaigns can enter the news system. The operation was linked to Iran, according to OpenAI, which said it had disrupted two sophisticated influence campaigns connected to Russia and Iran.
The Iranian network reportedly used seven fake journalist identities to produce articles that were published or syndicated by real news websites. The case shows how people running influence campaigns can use artificial intelligence (AI) to create content that looks like ordinary journalism and spread it through established publishing platforms.
OpenAI also identified a Russia-linked operation that allegedly created a fake research center in Latin America. The organization appeared to hire people who did not know that Russian operators controlled it.
This raises questions about how news editors verify writers, check the organizations behind submitted articles and identify content created as part of a coordinated influence campaign.
How AI Helped Fake Journalists Get Nearly 100 Articles Published
OpenAI’s October 8 report describes how the Iranian operation used ChatGPT to prepare articles and contact real news editors. Instead of simply posting propaganda on social media, the operators tried to get their work published on established websites.
The group used seven fake journalist identities, each with a name and a background that made them appear to be Western writers. Some profiles claimed experience covering international politics, human rights, the Middle East or armed conflicts.
The operators used ChatGPT to review long articles, improve their wording and check whether the drafts met the submission requirements of specific publications. They also asked the AI tool to prepare emails pitching the articles to editors.
OpenAI identified nearly 100 articles published or syndicated under these fake bylines between July 2025 and October 2026. The articles appeared across roughly a dozen online outlets, mainly covering international affairs, geopolitics and the Middle East.
The operation also generated social media comments about the US-Iran conflict. However, OpenAI found little evidence that these comments attracted significant attention from genuine users.
How Fake Journalists Got Past Editorial Checks at Real News Websites
The case highlights a weakness in the way online publications accept outside contributions. News websites often receive articles from freelance writers and contributors who do not work directly for their editorial teams.
A convincing article, a professional email and a believable biography can make a submission appear legitimate. AI tools can help someone prepare all three, making it easier to contact several publications with content that matches their editorial requirements.
In this case, the operators did more than generate text. They used ChatGPT to refine their articles for particular outlets and prepare pitches for editors. This allowed them to approach publications as if they were dealing with ordinary freelance journalists.
Some of the fake identities also had social media accounts that supported their claimed backgrounds. These accounts could make the writers appear more established, even though their identities were being used as part of a coordinated influence campaign.
OpenAI did not establish that every publication followed the same process or explain precisely why each editor accepted the submissions. Therefore, it would be inaccurate to conclude that all the editors simply failed to recognise AI-generated writing.
The central problem was that the campaign used false identities and misleading affiliations to get political messaging into genuine publishing channels.
Nearly 100 Fake Articles Targeted Political Debate on the US-Iran Conflict
The Iranian operation focused mainly on the US-Iran conflict. Its articles discussed geopolitical issues and often reflected anti-war or progressive positions, according to OpenAI.
The purpose of such a campaign is different from publishing a false claim on an anonymous social media account. An article appearing on a real news website can reach readers who already trust that publication.
Once published, an article may also be shared through the outlet’s social media accounts or circulated by other websites. This can give the content a wider audience and make its original source harder to recognise.
OpenAI said some of the publications that carried the articles had substantial social media followings. That gave the campaign a potential route to audiences beyond those directly targeted by its fake journalist profiles.
However, publication does not automatically mean that an article changed readers’ opinions or influenced political decisions. OpenAI’s findings show that the campaign secured placements, but they do not establish the full effect those articles had on public opinion.
Russia-Linked Network Used a Fake Research Centre to Spread Its Message
The second operation described by OpenAI followed a different approach. Rather than relying mainly on fake journalist profiles, the Russia-linked network allegedly created a false research organisation called the Social Research Center in Latin America.
The organisation presented itself as a research platform focused on the Indian diaspora in Latin America and relations between India and countries in the region. According to OpenAI, the Russian operators appeared to control the organisation while using local employees who did not know who was behind it.
These employees reportedly conducted interviews with experts and commentators and prepared research papers in good faith. OpenAI identified more than 60 articles on the centre’s website, most of which appeared to be original work.
The research covered subjects including public opinion on the BRICS group and comparisons of Brazil’s economy under former president Jair Bolsonaro and President Luiz Inácio Lula da Silva. The operators also used AI to prepare internal reports, create social media content and document the organisation’s activities.
OpenAI rated this Russia-linked operation Category 5 on its six-level Influence Operation Breakout Scale. It described it as the first Category 5 operation it had disrupted since it began publishing reports on these activities.
The Iranian operation received a Category 4 rating for its article-publishing activity. The ratings measure the operations’ potential reach and ability to break into authentic audiences, rather than proving that they achieved their intended political goals.
OpenAI Bans Accounts Linked to Russian and Iranian Influence Campaigns
OpenAI said it banned the ChatGPT accounts associated with both operations. It also shared information about the Iranian campaign with relevant authorities and industry partners.
The company explained that the operators combined AI with traditional methods of deception. ChatGPT helped them produce and refine content, but the campaigns also relied on fake identities, misleading organisations and efforts to establish credibility with real audiences.
OpenAI’s findings do not mean that every article written with AI is propaganda or that every freelance journalist using AI is operating deceptively. The concern is the deliberate use of AI to support hidden identities and coordinated political influence.
The report also shows that AI-generated content does not need to go viral on social media to reach a wider audience. Getting an article published by an established outlet can provide a different route to readers.
News Editors Face New Challenges in Verifying Freelance Journalists
The case could push news organisations to review how they verify freelance contributors and outside submissions. Checking the quality of an article is important, but it may not be enough to establish whether its author is genuine or whether a hidden organisation is coordinating the work.
Editors may need to check contributor biographies, confirm professional histories, examine the organisations behind submissions and look for signs that several articles are part of a coordinated campaign. Publications covering international affairs and conflicts may face particular challenges because outside contributors often submit analysis on fast-changing events.
AI detection tools alone are unlikely to resolve the problem. They cannot reliably establish who wrote a piece, whether the author is using a false identity or whether the article is part of a wider influence operation. The more important question is whether the publication can verify the writer’s identity, sources and affiliations.
OpenAI’s report shows how AI can make an existing form of deception easier to carry out at scale. The technology helped the operators prepare articles and pitches, but the campaigns depended on a broader effort to create false credibility.