Iranian operatives used AI to plant fake stories in multiple US media outlets

Morteza Nikoubazl/NurPhoto/AP/File

Operatives in Iran used OpenAI models and fake personas to help plant news articles critical of the US war against Iran in multiple US media outlets, OpenAI said on Thursday.

About a dozen outlets around the world ran roughly 100 articles by seven fake journalists, OpenAI said in a blog post . The firm said it banned ChatGPT accounts that were prompting the AI tool in Farsi to tweak articles and editorial pitches, and generate social media comments.

Some of the AI-assisted articles had provocative titles, including, “Is America’s Iran Policy Being Privatized? The Jared Kushner Question,” and “The Iran War and the Collapse of Western Moral Credibility.”

It’s the latest example of AI agents playing a key role in the propaganda battles that coincide with the US and Israel’s war with Iran. Young Iranians have also used AI to make viral Lego-themed videos mocking US leaders, while President Donald Trump has posted AI-generated videos depicting US strikes on Iran.

None of the publications that ran the articles appear to be a major, mainstream news outlet, but some of them have sizable online followings. One of the outlets, Middle East Monitor, which is based in London, has almost two million followers on Facebook.

Getting the anti-war narratives in front of a US audience could count as a successful information operation for whoever was behind it. Who that is, and whether they’re affiliated with the Iranian government, is unclear, OpenAI said.

“The overall activity appeared consistent with a commercial actor running a for-hire influence campaign, but we are unable to identify the particular actor involved,” the AI firm wrote in its blog.

The article about Kushner and US foreign policy was published by the LA Progressive, which describes itself as an “independent, justice-driven publication” based in Los Angeles. CNN has requested comment from LA Progressive.

Daud Abdullah, Director of the Middle East Monitor, told CNN he will look into the articles OpenAI flagged and “take the appropriate action.”

After the articles published, the Iranian operatives asked ChatGPT to generate comments they could post on social media purporting to be people reacting to the articles, according to OpenAI.

“The users asked for comments on topics such as missile strikes, ceasefires, and diplomatic maneuvering, specifying outputs that portrayed the US as the aggressor, Israel as an unreliable ally of the US, and Iran as a defiant but resilient victim,” the blog says.

“It’s remarkable that the Iranians managed to land their content in real media outlets, expanding the potential reach of their campaign,” said Jessica Brandt, who ran the Office of the Director of National Intelligence’s now-shuttered Foreign Malign Influence Center from 2023 to 2025.

“We’re witnessing AI enable foreign adversaries to better hide their hand, building more sophisticated false fronts than they might otherwise be able to in order to launder their content through authentic channels, leveraging those channels for reach,” Brandt told CNN.

OpenAI’s findings show how AI is increasingly being used to shape perceptions of war. This spring, a false AI-assisted US intelligence report about a purported Chinese ship transporting nuclear components almost led the US military to board the ship, CNN previously reported .

莫尔泰扎·尼库巴兹尔/NurPhoto/美联社/文件

OpenAI 周四表示，伊朗特工利用 OpenAI 模型和虚假身份，帮助在多家美国媒体上发布批评美国对伊朗战争的新闻文章。

OpenAI在一篇博客文章中表示，全球约有十几家媒体刊登了七名虚假记者撰写的约100篇文章。该公司称，已封禁了使用ChatGPT工具以波斯语修改文章和社论内容，并生成社交媒体评论的账户。

一些人工智能辅助撰写的文章标题颇具挑衅性，例如“美国的伊朗政策是否正在私有化？贾里德·库什纳问题”和“伊朗战争与西方道德信誉的崩溃”。

这是人工智能代理在美以对伊战争期间的宣传战中扮演关键角色的最新例证。伊朗年轻人也利用人工智能制作了以乐高积木为主题的病毒式传播视频，嘲讽美国领导人；与此同时，美国总统唐纳德·特朗普也发布了人工智能生成的视频，描绘美国对伊朗的空袭。

刊登这些文章的媒体似乎都不是主流新闻机构，但其中一些拥有相当数量的网络粉丝。例如，总部位于伦敦的“中东观察”（Middle East Monitor）在Facebook上拥有近两百万粉丝。

OpenAI表示，将反战叙事传递给美国民众，对幕后操纵者而言，可能算得上是一次成功的舆论战。但幕后操纵者是谁，以及他们是否与伊朗政府有关联，目前尚不清楚。

“整体活动似乎与商业机构开展的雇佣影响力活动相符，但我们无法确定具体是哪位参与者，”这家人工智能公司在其博客中写道。

这篇关于库什纳和美国外交政策的文章发表于《洛杉矶进步报》（LA Progressive），该报自称是一家总部位于洛杉矶的“独立、以正义为导向的出版物”。CNN已向《洛杉矶进步报》请求置评。

中东观察组织主任达乌德·阿卜杜拉告诉 CNN，他将调查 OpenAI 标记的文章并“采取适当的行动”。

据 OpenAI 称，文章发表后，伊朗特工要求 ChatGPT 生成评论，以便发布到社交媒体上，假装是人们对文章的反应。

该博客称：“用户要求就导弹袭击、停火和外交斡旋等话题发表评论，并特别要求发表将美国描绘成侵略者、以色列描绘成美国不可靠的盟友、伊朗描绘成不屈不挠但坚韧不拔的受害者的评论。”

“伊朗人能够将他们的内容投放到真正的媒体上，这令人瞩目，也扩大了他们宣传活动的潜在影响范围，”杰西卡·布兰特说道，她曾在2023年至2025年期间担任美国国家情报总监办公室现已关闭的外国恶意影响中心主任。

布兰特告诉 CNN：“我们正在目睹人工智能使外国敌对势力能够更好地隐藏其真实意图，建立比他们原本可能拥有的更复杂的虚假幌子，以便通过真实渠道洗白其内容，并利用这些渠道扩大影响力。”

OpenAI 的研究结果表明，人工智能正越来越多地被用于塑造人们对战争的认知。据 CNN 此前报道，今年春天，一份由人工智能辅助生成的关于一艘据称运输核部件的中国船只的虚假美国情报报告，几乎导致美军登船作战。