OpenAI Uncovers Russian and Iranian AI-Driven Influence Campaigns Targeting Western Media

OpenAI announced Thursday that it had dismantled two covert influence operations linked to Russia and Iran that exploited its AI models to masquerade as journalists and think‑tank analysts, inserting fabricated stories into mainstream outlets.

According to the AI lab, the campaigns created multiple false personas that published articles, op‑eds and commentary through reputable news sites, using ChatGPT and related tools to generate text that mimicked professional reporting. The fabricated output was distributed across a range of outlets, from niche blogs to internationally recognized newspapers, and the operations also set up front think‑tank entities that supplied research‑style pieces to editors, blurring the line between genuine analysis and state‑driven propaganda.

OpenAI said the groups were identified through internal monitoring and collaboration with external partners. The Russian network targeted topics such as NATO expansion and energy policy, while the Iranian effort focused on human‑rights narratives and regional security. Both aimed to shape public opinion in the United States and Europe by embedding their narratives within the normal news cycle. OpenAI’s security team flagged anomalous usage patterns, prompting a joint investigation with cybersecurity firms that traced the activity back to the two state‑linked operators.

The discovery highlights growing concerns among policymakers and technology firms about the misuse of generative AI for disinformation. Unlike traditional bot farms that rely on repetitive posting, large‑language models can produce nuanced, context‑aware content that passes editorial checks. Experts note that similar AI‑driven campaigns have been suspected in previous election cycles, raising alarms that the technology could amplify propaganda at scale. They warn that the ease of creating credible‑looking bylines could erode trust in journalism unless detection tools keep pace.

OpenAI indicated that it has blocked the accounts involved and is enhancing its usage‑policy enforcement to prevent similar abuse. The company also called for broader industry cooperation and for governments to consider regulatory frameworks that address AI‑enabled influence operations. OpenAI plans to roll out watermarking for AI‑generated text to aid publishers in distinguishing machine‑authored content. The incident underscores the need for newsrooms to verify author identities and for readers to remain vigilant about the sources behind online commentary.