OpenAI has identified a previously unreported relationship between two Iranian influence operations, STORM-2035 and IUVM, based on cross-platform activity. One banned ChatGPT account was found to generate content for both.

Four of the five ChatGPT accounts generated content published by entities linked to STORM-2035. One account generated long-form articles for al-sarira[.]com and tweets for its X account, as well as other X accounts. Another account generated Spanish-language articles for lalinearoja[.]net. Notably, a fifth account generated French-language texts for critiquepolitique[.]com (linked to STORM-2035) and English-language articles for iuvmpress[.]co (linked to IUVM). The operator of this fifth account appears to have rewritten model-generated text to evade detection.

The content produced was pro-Palestinian, pro-Hamas, and pro-Iran, while opposing Israel and the United States. It also praised Bashar al-Assad during Syria's political upheaval. Some banned accounts also used OpenAI's models for English and Spanish language teaching materials, a tactic associated with previous Iranian threat activity.

The influence operation had minimal impact, with limited social media following and engagement. OpenAI assesses its impact as low-end Category 2 on the Breakout Scale, indicating multi-platform activity without significant real-world traction.