OpenAI Disrupts Iran-linked Influence Operation IUVM

OpenAI has banned accounts associated with the Iran-origin operation known as IUVM. This network used AI models to generate and translate content for websites promoting pro-Iran, anti-Israel, and anti-US narratives.

Technical Behavior of the IUVM Campaign

IUVM utilized AI models to generate, proofread, and translate long-form articles, headlines, and website tags. The operation focused on targeting global audiences, specifically producing content in English and French.

Content creation followed a specific pattern: articles were typically generated the day before publication, while website tags were created immediately before publication, suggesting a high degree of automation. In one instance, OpenAI identified website tags that accidentally included the AI model's own response message, indicating either automation or a lack of proofreading.

Content Focus and Narratives

The influence operation focused on the production of long-form articles and headlines that praised Iran, the Palestinians, and the "Axis of Resistance," while maintaining an anti-US and anti-Israel stance. This content was published on the current IUVM website, iuvmpress.co, and other associated domains.

Impact Assessment and Scale

Using the Breakout Scale—a framework for measuring the impact of influence operations (IO) that rates activity from 1 (lowest) to 6 (highest)—OpenAI assessed the IUVM operation as a Category 2 operation. This classification indicates that while the network was posting activity across multiple platforms, it failed to achieve "breakout" or significant audience engagement on any of them.

Indicators of Compromise

OpenAI identified the following domains as being associated with the IUVM campaign:

• iuvmpress.co
• iuvmarchive.org