OpenAI has banned a cluster of accounts it says were likely operating out of China to flood the internet with AI-generated propaganda.
The company pulled the accounts as part of what it internally labeled Operation Sponsored Discontent. OpenAI says the network used its tools to generate English-language social media posts alongside separate Spanish-language articles. The takedown groups both content streams under a single influence operation. OpenAI has not disclosed how many accounts were involved or where the material was distributed.
The specifics matter less than the pattern. Language models make it cheap to produce fluent propaganda in multiple languages at once, without hiring native speakers or writers. That shrinks the cost of running an influence campaign right as platforms are still building the tools to catch AI-written text.
Banning accounts after the fact is good hygiene, but it doesn't undo whatever those posts already reached before anyone noticed.