OpenAI has banned a cluster of accounts tied to Iranian influence operations that used its AI tools to write propaganda.
The company said the accounts were connected to two previously identified campaigns: the International Union of Virtual Media network, known as IUVM, and an operation tracked as STORM-2035. Operators used OpenAI's models to generate articles and social media posts pushing narratives aligned with Iranian state interests, then spread that content across multiple platforms. OpenAI's disclosure, published as part of a report on malicious uses of its tools, did not specify how many accounts were caught or how much content they generated before the ban.
This is what AI-enabled propaganda actually looks like in practice: not deepfaked video, but ordinary-sounding text produced faster and cheaper than a human writer could manage. For platforms and readers, the real shift isn't the message. It's the volume and speed at which convincing filler can now be assembled.
IUVM has been flagged by researchers for years as an Iranian influence vehicle long before generative AI existed; the tools changed, the playbook didn't.