OpenAI blocks two ChatGPT clusters aimed at US AI infrastructure debate

The fight over AI infrastructure is now part of the influence-operations map. On June 10, 2026, OpenAI said it banned two clusters of ChatGPT accounts likely originating from China after they used its models to generate comments and images aimed at US AI and technology policy debates.

In the report, OpenAI names the first cluster “Data Center Bandwagon.” Its content pushed the idea that AI data center buildouts were raising electricity prices for ordinary families. The second cluster, “Tech and Tariffs,” generated material criticizing US tariffs as an attempt to dominate technology competition. OpenAI says prompts in that cluster specified that outputs should exclude Xi Jinping and include only President Trump.

The second cluster was also connected to likely inauthentic social media accounts that targeted OpenAI with false claims that ChatGPT user data had been compromised. OpenAI says those breach allegations were entirely false, and it found no evidence that the influence activity achieved meaningful breakout beyond its own network.

That last point is important. The story is not that these campaigns successfully moved public opinion. The sharper signal is that operators are testing narratives around AI infrastructure, energy costs, tariffs, and trust in model providers. Those are already live political and economic debates in the US, which makes them useful hooks for covert amplification.

For AI companies, this is a security problem that sits between model abuse, platform integrity, and geopolitical risk. Blocking accounts is only one layer. Detection has to connect prompts, generated media, coordinated posting behavior, and false claims moving through external accounts. For policymakers, the report is a warning that AI governance debates can themselves become targets, especially when local concerns over power prices and data center construction are easy to repackage into broader anti-AI narratives.