OpenAI bans ChatGPT accounts tied to Russian, Iranian ops

OpenAI has banned ChatGPT accounts linked to two separate influence campaigns, one run from Russia and one from Iran. Both used the AI tool to support propaganda spread through social media, news outlets and, in the Russian case, a fake research organization.

The company described both operations in a safety report published on Thursday. It said the operators combined AI with older, more traditional influence tactics.

A fake think tank aimed at Latin America

The Russian campaign targeted audiences across Latin America. Its centerpiece was the Social Research Center (SRC), which presented itself as a research platform studying the Indian diaspora in Latin America and India's relations with countries in the region.

According to OpenAI, Russian operatives appeared to control the center through a fabricated identity. The campaign also recruited local people in Latin America to carry out research for the SRC, apparently without telling them who was behind it.

"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," OpenAI said.

Much of what the center published looked like original research written by these recruits, not AI-generated text, the report noted.

The main goal was to damage Ukraine's reputation in Latin America. Some activities, however, appeared aimed at influencing domestic politics in countries such as Argentina and Bolivia.

The operators used ChatGPT to produce fake leaked documents and scripts for audio recordings, some of which circulated online. At least some of the false claims spread far enough to trigger fact-checks and official denials.

OpenAI did not attribute the campaign to a named threat actor. It said evidence pointed to links with a Russian organization known as Politology or La Compania. That group has been reported to be a successor to influence operations tied to the late Russian oligarch Yevgeny Prigozhin and his Wagner Group, the private military company.

The operators mostly wrote to ChatGPT in Russian and appeared to be based in Russia.

"Since we do not allow access to our models from Russia, they used VPNs to connect to our services," OpenAI said.

Seven fake journalists pitching Iran stories

The Iranian operation worked differently. It relied on seven fake journalist personas to pitch articles about the U.S.-Iran conflict to online publications.

OpenAI found nearly 100 articles published or republished under these identities across roughly a dozen outlets. Most were small and midsize publications covering international affairs, geopolitics and the Middle East. The earliest article dates from July 2025, the latest from October 2026. OpenAI did not name any of the outlets.

ChatGPT was used to polish long articles, write pitches to editors and draft internal reports. The operators also generated batches of social media comments in English and Persian on political and geopolitical topics.

Those comments got little traction, with relatively few likes, views or replies. Some of the outlets that ran the articles had large audiences, though. One had nearly 2 million Facebook followers, according to OpenAI.

The accounts appeared to originate in Iran, mostly used Persian-language prompts and reached ChatGPT through VPNs. OpenAI did not identify the people or organizations responsible.

Old playbook, new tools

The report highlights that both campaigns went beyond fake social media profiles and used seemingly independent organizations to carry their messages.

"What is most striking about these operations is that they closely resembled complex influence operations of the pre-AI age, but used AI to make some of the workflows easier," OpenAI said.

Our Take

The most useful takeaway here is that AI did not change what these operations were. It changed how fast they could work. Front organizations, unwitting local recruits and fake bylines are long-established techniques. ChatGPT mainly helped with drafting, editing and producing content at volume.

This suggests that defenders and editors should not focus only on spotting machine-written text. In the Russian case, much of the published research was written by real people. Efforts such as OpenAI's textGrain watermarking may help in some cases, but they would not catch content that humans wrote in good faith.

For small and midsize publications, the Iranian campaign is a reminder to verify freelance contributors, especially those pitching on geopolitics.

It is also worth noting that geoblocking did little here, since both groups simply used VPNs. Together with recent concerns about AI agents abusing public platforms, it will be worth watching whether AI providers move toward stronger account-level controls rather than relying on location-based restrictions.