Iran ran a covert influence operation targeting American news outlets — and dozens of editors never knew they were being played.
OpenAI just exposed a campaign the company calls “Bogus Bylines” — seven fake journalist personas who used ChatGPT to polish antiwar articles and plant them across at least 20 US publications.
The revelation marks a troubling evolution in foreign influence operations. Unlike traditional disinformation campaigns that rely on bot networks and fake social media accounts pushing content from the outside, this operation successfully breached the editorial walls of legitimate American news organizations. The perpetrators didn’t just shout propaganda into the void — they got editors to publish it under the guise of authentic American commentary.
The Port St. Joe Star, a small Florida newspaper, published one such article in September under the byline Ervin B. Hoskins. The piece criticized the U.S.-Iran conflict and appeared to be a standard opinion submission.
But Hoskins doesn’t exist.
According to OpenAI’s report released Thursday, Hoskins was an Iranian-run fake persona. His author bio claimed he was “an American freelance writer,” but social media transparency settings traced his accounts back to Iran. OpenAI’s image-verification tool confirmed the author photo was AI-generated.
“This persona’s bio claimed to be ‘an American freelance writer,’ but social media transparency settings located the accounts in Iran.”
The operation used ChatGPT to refine English-language articles about the U.S.-Iran war and tailor submissions to meet specific editorial guidelines at targeted outlets. Operators also generated batches of social media comments supporting the articles — some posted within minutes of one another.
The sophistication of the approach is particularly concerning. Rather than producing obvious propaganda filled with grammatical errors and stilted phrasing that might raise red flags, the operators leveraged artificial intelligence to create polished, professionally written content virtually indistinguishable from legitimate freelance submissions. They studied each publication’s style, tone, and submission requirements, then used AI tools to ensure their pieces would pass initial editorial review.
OpenAI identified almost 100 articles published or syndicated under the fake bylines across more than a dozen publications. The Washington Post reported the campaign successfully infiltrated at least 20 news outlets.
The scale and duration of the operation suggests careful planning and sustained effort. This wasn’t a one-off experiment but a systematic campaign that operated for months, continuously adapting and expanding its reach across the American media landscape.
The earliest planted article appeared in July 2025. Publication frequency spiked after the U.S.-Iran war broke out earlier this year.
The Middle East Monitor alone published at least 21 articles under fake bylines, including pieces attributed to Hoskins, Michael Harrison, Jenny Williams, and Alice Johnson — all fabricated personas.
Each fake persona was carefully constructed with its own author bio, social media presence, and writing style to create an illusion of legitimacy. The operators clearly understood that editors often rely on these digital credentials when vetting unfamiliar contributors, especially for opinion and analysis pieces that don’t require the same rigorous fact-checking as investigative reporting.
OpenAI said the activity appeared consistent with a commercial operator running the campaign for hire, but the company could not identify who funded it.
The involvement of a commercial operator suggests this type of influence campaign may be available as a service to state actors and other entities willing to pay. This commodification of disinformation represents a dangerous new frontier where sophisticated propaganda operations can be outsourced to skilled practitioners who have perfected the craft of deceiving American editors.
While the operation’s social media comments attracted little traction, the articles themselves reached substantial audiences. One publication that carried the fake journalism had almost 2 million Facebook followers, 355,000 followers on X, and over 544,000 on Instagram as of August 2026.
The disparity between the failed social media strategy and the successful article placement reveals an important insight: foreign actors have learned that the credibility conferred by legitimate news organizations is far more valuable than raw social media reach. A single article published by a real newspaper carries exponentially more persuasive weight than thousands of bot-generated tweets.
The articles typically focused on international affairs, geopolitics, and Middle East events — often from an antiwar or progressive angle designed to erode American resolve.
The content strategy was clearly designed to exploit existing political divisions within American society. By framing arguments in terms familiar to certain ideological perspectives, the operators increased the likelihood that editors sympathetic to antiwar viewpoints would accept the submissions as genuine expressions of American dissent rather than foreign manipulation.
OpenAI’s report raises uncomfortable questions about editorial vetting at outlets that pride themselves on fact-checking and transparency.
The mullahs just proved they can weaponize America’s free press without firing a shot.
