OpenAI Reveals How Iranian Operatives Used AI to Spread Anti-US War Narratives

Iran-linked operatives allegedly used artificial intelligence to produce misleading articles and social media content critical of the United States during the conflict involving Iran, according to an investigation released by OpenAI. The findings raise fresh concerns about the growing use of AI technology in political influence campaigns and the spread of deceptive information during international conflicts.

According to OpenAI, the operation involved fabricated journalist identities, AI-assisted article writing and coordinated online messaging intended to influence public perceptions of US foreign policy. The content reportedly presented Washington as an aggressor while portraying Iran as a country resisting external pressure. The investigation highlights how AI tools can be misused to create and distribute political narratives across different platforms and languages, potentially making it harder for readers to distinguish authentic reporting from organised influence efforts.

OpenAI said the individuals involved used ChatGPT to develop and refine long-form articles, prepare submissions to online publications and draft emails to editors. The operation reportedly relied on seven fictional journalist profiles that presented their owners as Western reporters or commentators specialising in international affairs, human rights and geopolitics. The company identified nearly 100 articles published or syndicated through approximately a dozen online outlets worldwide. The earliest article identified during the investigation dated back to July 2025, while the campaign's activity reportedly increased as the US-Iran conflict intensified in 2026. However, OpenAI said it could not establish the precise identity of the organisation behind the operation or confirm direct control by the Iranian government.

The articles reportedly addressed military operations, diplomatic negotiations, ceasefire discussions and the wider consequences of US involvement in the conflict. Their messaging frequently questioned Washington's motives, criticised Israel's relationship with the United States and attempted to present Iran's position in a more favourable light. The material was produced in English and Persian, allowing the operators to target audiences across different linguistic communities.

Beyond publishing articles, the campaign allegedly used AI to generate groups of social media comments designed to support its narratives. Some comments were written to resemble ordinary reactions from readers, creating the appearance of public approval for particular articles and political arguments. Such tactics can make a coordinated operation seem like a broader public movement, even when the accounts involved represent only a small group of operators.

OpenAI reported that the social media activity it identified generally received limited engagement. Nevertheless, the publication of articles through established websites presented a different challenge because readers may place greater trust in material that appears on familiar news platforms than in anonymous social media posts. The findings also demonstrate why verifying the identity and background of contributors has become increasingly important for media organisations. A fabricated journalist profile can be used to approach editors, submit apparently professional commentary and distribute politically motivated material without immediately revealing the author's real affiliations. If the content is published, it may reach audiences who have no reason to suspect that the contributor's identity or intentions have been misrepresented.

Criticism of American foreign policy should not automatically be classified as disinformation. Opposition to military action, support for ceasefire negotiations and criticism of government decisions are legitimate parts of public debate. The key concern in the reported operation is the alleged use of deceptive identities and concealed coordination to promote particular narratives while disguising their origins. Artificial intelligence can make such activities easier to organise by helping users write articles, translate content, develop multiple versions of a message and prepare material for different platforms. Tasks that once required considerable time and editorial effort can now be completed more quickly. However, producing large amounts of content does not necessarily mean a campaign will persuade readers or significantly change public opinion.

OpenAI said it banned accounts associated with the activity and shared information about the operation with relevant authorities and industry partners. The company has previously reported other attempts to misuse its AI tools for covert political messaging, including operations involving content about elections and international conflicts. These cases suggest that generative AI is increasingly being incorporated into established propaganda techniques rather than replacing them entirely.

The incident reinforces the importance of checking contributors' identities, examining the credibility of sources and reviewing unusual patterns in unsolicited submissions. Readers can also protect themselves by comparing reports across reputable publications, checking whether claims are supported by verifiable evidence and distinguishing factual reporting from political commentary. AI-detection software alone is unlikely to provide a reliable solution because AI-assisted writing can resemble ordinary human writing, while genuine human-written material can also be edited using AI tools. Establishing the credibility of a story therefore requires a broader assessment of its sources, authorship, evidence and publication history.

OpenAI's findings underline a growing challenge for the global information environment. As governments, political groups and other actors explore the capabilities of artificial intelligence, the technology can be used both to improve access to information and to manufacture misleading narratives at scale. The Iranian-linked operation described in the investigation illustrates how fabricated identities and AI-generated material can be combined in attempts to influence perceptions of a war.

Although the full impact of the reported campaign remains uncertain, the case highlights the need for greater transparency in online publishing and stronger verification practices across the news industry. In an era when convincing content can be produced quickly and distributed internationally, understanding who created a message, what evidence supports it and why it is being circulated is becoming essential to maintaining trust in digital journalism.