On Thursday, October 8, 2026, OpenAI published a report on its blog explaining that it had banned Russian and Iranian accounts that were using its artificial intelligence tools to support influence operations in several countries. According to NPR’s reporting, the operatives themselves claimed in internal documents that they had managed to get false narratives picked up by news outlets and, in one case, to prompt a response from a member of the European Parliament.
The most striking aspect of the Russian case is how ChatGPT was used. According to OpenAI, the accounts were mostly writing reports for unspecified superiors that detailed their work. That practice, meant to save bureaucratic effort, ended up giving the company an unusual view of campaigns that had not previously been attributed to Russia. Generating fake material with AI was, the company said, only a minority of the workload.
The Russian operations targeted countries in Latin America and appeared aimed at undermining support for Ukraine and influencing local politics. OpenAI said one of those reports appeared to describe the widest-reaching influence operation the company has come across so far, precisely because it led to a public response from a political figure.
According to the operatives’ own account, the group impersonated a regional educational authority in Peru and tricked some schools into organizing activities about Ukraine that featured Stepan Bandera. Bandera is a controversial figure: admired by nationalists for leading a drive for Ukrainian independence before and during World War II, he is reviled by others, including in Poland, because his movement sided with the Nazis and killed Poles and Jews. Peruvian outlet Extra reported that one such event did take place at a school.
The chain did not stop in Peru. A Polish news report about the school event prompted a far-right Polish member of the European Parliament to suggest banning the entry of people who professed support for Bandera. “The Russian fake thus both fed on, and fed into, historical tensions between Ukrainians and Poles,” OpenAI wrote in its report, as quoted by NPR.
The internal memos produced with ChatGPT also claimed that the operatives ran a fake think tank in Latin America and had recruited unwitting researchers to interview experts and compose reports. OpenAI said that effort was similar to a 2020 Russian operation that posed as an independent news outlet linked to the late oligarch Yevgeny Prigozhin’s Internet Research Agency, which was involved in efforts to influence the 2016 U.S. election.
OpenAI added an important caveat: in other instances, the Russian operatives took credit for events they were not responsible for. In other words, the internal reports are not necessarily a faithful record of what they achieved, but also a way of presenting results to their bosses. The company investigated the impact of the various campaigns and concluded that those focused on placing false narratives in established outlets appeared to have reached a wider audience than those relying on fake social media accounts.
Darren Linvill, co-director of Clemson University’s Watt Family Innovation Center Media Forensics Hub, told NPR that Russians are constantly engaging in these sorts of “active measures”: narrative laundering, fake stories and invented personas to create or spread propaganda aligned with Russia’s worldview and political agenda. On the use of AI, he summed it up: the Russian underlings were trying to use it as a shortcut to do less work, as many people do, “but in this case it helped get them caught.”
The Iranian operation worked differently. Its accounts used AI to produce content in Persian and English, focused mainly on the U.S.-Iran war and, to a lesser degree, on American politics. OpenAI said the activity “appeared consistent with a commercial actor running a for-hire influence campaign,” although it was unable to identify who was behind it.
According to the company, the Iranian accounts created seven fictitious authors who pitched and polished long articles for small outlets around the world, publishing almost one hundred pieces this way. The outlets included the community section of the progressive U.S. site Daily Kos and the U.K.-based Middle East Monitor. According to NBC News, Middle East Monitor published at least 21 articles by four of those fake authors.
Daily Kos told NPR that its community section, where one of the fake personas posted, is separate from its staff reporting: anyone can sign up and posts are not reviewed before they appear, although the rules prohibit pretending to be someone else. For Middle East Monitor, Nasim Ahmed wrote that increasingly sophisticated AI-enabled influence operations present challenges across the industry, and that the outlet is reviewing its submissions process and strengthening its editorial safeguards.
For Linvill, AI’s value to the Iranians went beyond saving time: it helped them create language that looked and read as authentic, to the point that legitimate outlets published it. According to the researcher, a few years ago an Iranian operation of this kind would have struggled much more to pull that off. It is a direct warning to under-resourced newsrooms that accept outside contributions.
The case illustrates both sides of AI in the disinformation arena. On one hand, models make it easier to write fluently in other languages and to produce believable texts at scale. On the other, the fact that these actors use commercial services leaves traces that companies can detect and analyze. Reports like this one also give researchers and journalists leads to identify campaigns that would otherwise go unnoticed.
In short, OpenAI has documented how Russian operatives used ChatGPT almost as an office tool to report on campaigns in Latin America, while an Iranian network used it to fabricate authors and place articles in real outlets. The figures and successes described in the internal reports should be read with caution, because the operatives themselves exaggerated, but the episode confirms that vetting contributors and sources has become a central task for any media organization.
