OpenAI bans accounts used in influence operations

In an October 8 report, OpenAI said it banned accounts linked to two influence operations originating in Russia and Iran. According to the company, they hid their origins behind fictitious entities and used ChatGPT to prepare content and reports.

The Iranian operation included seven journalist personas and generated social media comments. The report describes the banned accounts’ use; it does not establish that all activity by those organizations has ended.

Interpreting the report’s reach

The Breakout Scale, published by Brookings, classifies operations by their spread across communities, platforms, media and public figures, through to possible offline actions. It describes how far an operation spread.

That classification is not a measurement of impressions, clicks or people persuaded. A fake persona cited by a news outlet and an AI-generated comment are different pieces of evidence. Keep that distinction when reading the cases: what was generated, where it appeared and what spread could be observed.