OpenAI: Iranian influence operation published 100 articles with 7 fake journalists

The American company OpenAI announced on Thursday, 8 October 2026, that it had halted an influence operation originating in Iran that created seven fake Western journalists and used ChatGPT to publish nearly 100 articles on about 12 online outlets around the world. Most of the articles dealt with the Iran-US war and took an anti-American line.

According to Radio Free Europe and CNN, these fake identities presented themselves as American or Western freelance journalists and had accounts and AI-generated images on social media. One of them, under the name "Erwin B. Haskins", had published dozens of articles criticising the US war against the Islamic Republic, even though no such person exists. OpenAI blocked the related accounts.

How the Iranian influence operation worked

According to OpenAI, the operators used prompts in Persian to ask ChatGPT to edit English drafts, match their tone to each publication's rules and write pitch emails to editors. The publication of the articles began in June–July 2025 and rose sharply after the war began.

According to Iran International, after the articles appeared the group generated batches of fake comments to amplify them, and had also built more than two dozen batches of hostile Persian-language replies to attack posts by Iran International, one of the independent Persian-language media. OpenAI was unable to identify the operators precisely and said their activity resembled a commercial influence-operation contractor; its direct link to the government has not been established. CNN wrote that none of the outlets were major media, and that one of them removed the articles after the matter came to light.

The operators used Persian-language prompts in ChatGPT to write articles under the names of fictitious American journalists.

Use of AI to identify military targets

This is not the first case. According to Iran International, the company Anthropic announced in September that it had halted the activity of an Iran-linked group that used the AI "Claude" to gather public information and prepare targeting proposals against US naval forces in the region, including building a programme to track the positions of carriers, identifying US military personnel from the captions of public photos and examining the vulnerabilities of ships' communication systems. According to Radio Free Europe, Anthropic also said Islamic Republic government bodies had used these tools for "cognitive warfare" campaigns and for monitoring ethnic minorities and opponents.

The revelations come as the Islamic Republic at home punishes citizens for posting protest content on social media with death sentences and, abroad, according to experts, tries with targeted propaganda to shape Western public opinion about the Iran-US war.

Sources

• Iran International
• RFE/RL
• CNN