In an October 8, 2026 threat report, OpenAI said it banned a cluster of ChatGPT accounts behind an operation it nicknamed "Bogus Bylines" — seven fabricated Western journalists that pitched and placed roughly 100 articles, mostly critical of US military action against Iran, across about a dozen small-to-medium online outlets. OpenAI called the activity commercial in nature and did not attribute it to the Iranian government directly.
2026-10-10 · by Moe Ameen
On October 8, 2026, OpenAI published a threat-intelligence report disclosing that it had banned a cluster of ChatGPT accounts tied to an Iranian-linked influence operation it nicknamed "Bogus Bylines." The operators invented seven fake Western journalists — most posing as Americans writing about international politics, human rights, and the Middle East — and used ChatGPT to draft and polish articles, tailor each piece to a publication's submission requirements, and write the emails pitching the stories to editors. OpenAI said roughly 100 articles were published or syndicated under those fabricated bylines across about a dozen small-to-medium online outlets focused on international affairs and the Middle East.
The content was largely English-language commentary critical of US military action against Iran, with operators prompting ChatGPT mostly in Persian and asking for copy that cast the US as the aggressor. Publication under the fake personas began in July 2025 and continued through October 2026, and the volume rose sharply after the US-Iran conflict escalated. The operators also used ChatGPT to generate batches of hostile social media comments; OpenAI reported no evidence of significant public engagement with those. OpenAI rated the operation Category 4 on its six-point Influence Operations "Breakout Scale," below the Category 5 it assigned a separate Russian operation disclosed in the same report.
OpenAI stopped short of attributing the campaign to the Iranian government directly. It described the activity as appearing commercial in nature — consistent with a hire-for-service influence operation — and said the exact affiliations, including any government links, remain unconfirmed. The specific outlets that ran or republished the content have been named in subsequent press coverage rather than by OpenAI, whose own description kept the outlets generic; several outlets reportedly removed the fake author pages afterward. Treat the exact article counts and outlet lists as OpenAI's and reporters' figures rather than independently audited numbers.
The useful lesson here is not "AI writes articles now" — it is that the entire problem was deception: fabricated journalists, no disclosure, and no real person accountable for a word of it. AI-assisted content is not the liability; faking an identity is. The legitimate way to run a model at scale is the mirror image of Bogus Bylines — you generate under your own brand, in your own voice, and a human signs off on what ships. That is the shape [Kompozy](/) is built around. A [Persona Brief](/glossary/persona-brief) encodes your actual voice and guardrails so output sounds like you rather than a generic byline, and every piece passes a per-post review gate before it posts — the accountability step the operation deliberately skipped.
From there it is distribution you own, not credibility you borrow. Kompozy generates the full mix — short and avatar videos, carousels, quote graphics, text posts, a blog, and a newsletter — and schedules it across the eight social platforms plus blog and email from one queue, with [Autopilot](/glossary/autopilot) keeping a steady cadence running behind that review gate. The influence op needed other outlets' audiences because it had none of its own; a real creator's answer is to build a transparent, consistent presence on their own channels, where the voice is theirs and the posting is disclosed. Same AI efficiency the operators abused, pointed at an honest goal: an on-brand audience that knows exactly who it is hearing from. For related ground, see [AI-generated fake experts quoted in major media](/news/ai-generated-fake-experts-quoted-in-major-media), [Google's warning on fake AI author attribution](/news/google-warns-fake-ai-author-attribution), and [LinkedIn's inauthentic-activity enforcement](/news/linkedin-inauthentic-activity-enforcement).
It was an Iranian-linked influence campaign OpenAI disclosed on October 8, 2026. Operators used ChatGPT and seven fabricated Western-journalist personas to write and pitch articles — roughly 100 of them, mostly critical of US military action against Iran — to about a dozen small-to-medium online publications, then had the pieces published under those fake bylines.
Not directly. OpenAI said the activity appeared commercial in nature, consistent with a hire-for-service operation, and that the exact affiliations, including any government links, remain unconfirmed. Press coverage describes the campaign as Iranian-linked rather than confirmed state-run.
To draft and polish long-form articles, tailor each piece to a publication’s submission requirements, write the emails pitching the stories to editors, and generate batches of hostile social media comments. OpenAI said it found no evidence of significant public engagement with the comments.
The problem was deception and fake identities, not AI writing itself. Producing content under your own disclosed brand, in a consistent voice, with a human reviewing each post before it ships, is the legitimate pattern — and it is how content engines like Kompozy are designed to work.