OpenAI Says AI-Enabled False Fronts Are Reaching Real Media

OpenAI says it banned two covert influence operations—one originating in Russia and another in Iran—that used its models alongside fake organizations, fabricated personas and traditional media tactics. The disclosure shows how generative AI can make deceptive campaigns more efficient without replacing the human operators who build and direct them.

In a report published October 8, OpenAI said the Russia-origin operation received its highest rating yet under the company’s six-point Influence Operations Breakout Scale, scoring 5. The Iran-origin operation scored 4. OpenAI said both campaigns used “false front” entities to launder geopolitical or conflict-related messaging to audiences that might not know who was behind it.

How the campaigns worked

The Iranian operation used seven different bylines to pitch long-form articles to online outlets around the world. OpenAI identified almost 100 articles published or syndicated across more than a dozen publications, generally focused on international affairs, geopolitics and the conflict between the United States and Iran. The operators used ChatGPT to refine articles and editorial pitches, generate social media comments and prepare internal reports.

The Russian operation appears to have co-opted unwitting people in Latin America to run a think tank on the ground. OpenAI said the operators also created fake leaked documents and audio scripts, some of which were spread online, and used AI extensively to draft internal reports. The available evidence indicated that people working for the operation did not necessarily know who controlled it.

OpenAI said both campaigns managed to place content in mainstream media outlets rather than relying only on accounts controlled by the operators. It said this pattern has appeared repeatedly in its investigations of 30 covert influence operations over the past two and a half years.

Why the disclosure matters

The report does not show that AI alone created a mass audience for either campaign. Instead, it shows how AI can reduce the cost of routine influence work, including drafting, editing, translation, persona management and the production of large volumes of tailored content. Human operators still supplied the objectives, infrastructure and distribution strategy.

That combination creates a verification problem for publishers, officials and ordinary readers. A website can appear to have research credentials, a person can appear to be an independent journalist, and an article can be carefully edited for a publication without revealing who commissioned it.

OpenAI said the operations resembled influence campaigns from before generative AI, but used AI to make some workflows easier. The company said the campaigns’ covert nature also makes them vulnerable to responsible disclosure, because exposing the people and entities involved can make it harder for the operations to continue.

What organizations can do

1. Check who owns and funds a research group, publication or advocacy organization before relying on its material.
2. Verify authors, citations and methodology independently rather than treating a polished website as evidence of credibility.
3. Look for coordinated activity across multiple platforms, especially when accounts promote the same claims or direct readers to a newly established authority.
4. Require human review for political, security and geopolitical content produced or edited with generative AI.

OpenAI’s report is a reminder that serious AI-enabled influence campaigns may not look like obvious bot activity. They may resemble ordinary research, journalism or community outreach until investigators connect the people, websites and distribution channels behind them.

References

1. OpenAI. (2026, October 8). Disrupting AI-enabled “false front” operations. https://openai.com/index/disrupting-ai-enabled-false-front-operations/