OpenAI Exposes Iran Operation Using Seven Fake Journalists
OpenAI banned accounts behind a campaign that slipped war commentary into real media under fabricated bylines. Its investigation found almost 100 published or syndicated articles.
OpenAI said on October 8, 2026 that it had disrupted an influence operation originating in Iran which used seven fabricated journalist identities to place articles about the US war with Iran in online media. The company banned the associated ChatGPT accounts and shared information with relevant authorities.
Seven identities, real publications
In its report, Disrupting AI-enabled “false front” operations, OpenAI called the campaign “Bogus Bylines.” Investigators identified almost 100 articles published or syndicated under its bylines across roughly a dozen outlets worldwide. The publications were generally small or medium outlets covering international affairs and the Middle East.
The seven personas were Ervin B. Hoskins, Noah Lamington, Sophia Gonzalez, Michael Harrison, Ericka Feusier, Jenny Williams and Alice Johnson. Their biographies presented them as Western journalists, usually American; Lamington’s claimed a New Zealand location. These are the fabricated identities named by OpenAI, not allegations against genuine journalists with similar names.
The earliest article OpenAI identified appeared in July 2025 and the latest in October 2026. Publication frequency increased after the conflict began, with the report’s timeline showing a concentration from March 2026.
How the pitches reached editors
Operators submitted English drafts to ChatGPT for review against particular outlets’ submission requirements. After approving revisions, they requested emails pitching the articles to editors. The tool also helped refine internal reports and, in one case, write a persona’s biography.
Some identities had supporting social accounts. OpenAI said at least some published articles were subsequently promoted by the outlets themselves. That gave the campaign access to established audiences, rather than only followers of accounts controlled by its operators.
OpenAI traced the ChatGPT accounts to Iran and said the activity appeared consistent with a commercial influence campaign for hire, but it could not identify the specific operator. The disclosure did not establish Iranian government direction.
War narratives and weak comment engagement
The operators also requested comments in English and Persian portraying the United States as an aggressor, Israel as an unreliable American ally and Iran as a resilient victim. They generated more than two dozen batches of hostile Persian replies targeting Iran International, a broadcaster critical of Iran’s rulers.
OpenAI found little substantial engagement with the comments it located. Internal impact reports counted views on the original posts, rather than the replies, exaggerating apparent reach. It assessed the article placement activity at Category 4 on the six level Breakout Scale, while the commenting activity reached Category 2.
An earlier campaign, a different route
On August 16, 2024, OpenAI disclosed Storm-2035, an Iranian influence operation that generated political articles for five websites posing as news outlets, alongside social comments. At that time, it found no meaningful audience engagement.
The new case highlights a different entry point: fabricated contributors approaching genuine editors. Readers encountered the resulting articles through real publications, with bylines that concealed their origin. The verified publication count demonstrates distribution, not how many readers were persuaded.
