OpenAI Bans China-Linked Accounts Over Influence Campaigns
OpenAI bans China-linked accounts that allegedly used ChatGPT to create social media content aimed at influencing public opinion. The company revealed the findings in a report released on Wednesday. According to OpenAI, the accounts generated posts about AI data centers, electricity costs, and U.S. trade policies.
The report stated that a group of ChatGPT accounts from China produced content claiming AI data center expansion raised electricity prices for American households. OpenAI said the accounts were likely connected to a social media operations team at a private Chinese technology company. The firm reportedly worked with provincial-level government clients in China.
OpenAI bans China-linked accounts that also created content criticizing U.S. tariffs. The generated posts described tariffs as efforts to dominate technological competition. OpenAI found prompts directing writers to mention President Donald Trump while excluding references to Chinese President Xi Jinping.
The company said the campaign appeared designed to influence international audiences. However, OpenAI could not confirm any direct institutional affiliation.
According to the report, another network spread claims that ChatGPT user data had been compromised. OpenAI rejected those allegations and described them as completely false.
Key findings included:
- Content targeting AI data center development.
- Posts criticizing U.S. tariff policies.
- Messages focused on economic concerns.
- Claims regarding ChatGPT data security.
- Activity linked to likely inauthentic social media accounts.
OpenAI bans China-linked accounts as part of its efforts to prevent platform misuse. The company noted that most identified posts received little engagement. Despite that, OpenAI called the findings significant because they revealed tested narratives and influence strategies.
The company also described it as remarkable that American AI tools generated content about American AI policies. OpenAI added that its services remain unavailable in China, and the accounts accessed the platform through VPNs.
China's embassy in the United States told Reuters it was unfamiliar with the report and opposed what it called groundless attacks or smears against China.