Executive Summary
OpenAI has identified and banned two clusters of ChatGPT accounts linked to a covert influence operation likely originating from China. These campaigns, named "Data Center Bandwagon" and "Tech and Tariffs," used OpenAI's models to generate social media content aimed at manipulating US public debate on AI infrastructure and technology policy. While OpenAI found no evidence of significant public impact, the company is publishing its findings to help industry and government partners better identify and disrupt such threats.
Key Takeaways
* Two Campaigns Identified: OpenAI disrupted two distinct operations: "Data Center Bandwagon" and "Tech and Tariffs."
* "Data Center Bandwagon": This cluster generated content claiming that AI data center construction was causing electricity price hikes for American families.
* "Tech and Tariffs": This cluster created comments criticizing US tariffs, specifically mentioning President Trump while avoiding mention of China's leader. It also spread false allegations of a ChatGPT user data compromise.
* Methodology: The actors used ChatGPT to generate text and images for social media posts as part of their influence operations.
* Limited Impact: The report states there was no evidence of meaningful public engagement or breakout beyond the operation's own inauthentic activity.
* Tactic: The campaigns attempted to co-opt and amplify existing public concerns (e.g., energy prices) to build credibility and sow distrust.
Strategic Importance
This report highlights the emerging threat of state-sponsored actors using generative AI for covert influence and demonstrates OpenAI's proactive effort to police its platform. It serves as a public warning about how foreign adversaries are testing AI tools to manipulate democratic discourse and target critical technology infrastructure.