OpenAI Says Likely China-Linked Campaigns Used ChatGPT to Target U.S. Debates
Summary
OpenAI says two likely China-linked influence clusters used ChatGPT to create anti-data-center and anti-tariff content while posing as Americans online. The campaigns showed little meaningful engagement, and OpenAI did not attribute them directly to China’s government.
Key points
- OpenAI identified two activity clusters it says likely originated in China; it did not directly attribute them to the Chinese government.
- The “Data Center Bandwagon” cluster used ChatGPT to create content claiming data centers were raising U.S. electricity prices.
- A second cluster used ChatGPT to portray tariffs as a means of controlling global technology and to create comments and comics in multiple languages.
- Operators reportedly used VPNs, prompted ChatGPT in simplified Chinese, and posed as Americans on X and YouTube.
- OpenAI linked the anti-data-center content to an unnamed Chinese technology company with contracts with regional Chinese governments.
- OpenAI reported little meaningful engagement from targeted audiences and said it saw no signs the campaigns succeeded.
- The actors also used ChatGPT to edit reports containing campaign operational-security details and to discuss Meta’s content systems and detection evasion.
Article Details
- Event Type
- Two likely China-linked influence activity clusters used ChatGPT to generate content aimed at inflaming U.S. debates about data centers and tariffs.
- Impact
- The campaigns posed as Americans on social media and sought to expand their reach, but OpenAI reported little evidence of engagement beyond their own amplification networks and no signs of meaningful success among targeted audiences.
People
Vendors
Products
ChatGPTOpenAI’s threat intelligence team identified two likely China-linked influence activity clusters that used ChatGPT to generate anti-data center and anti-tariff content aimed at inflaming U.S.FacebookAnother report fed into ChatGPT discussed how best to leverage Facebook’s content ecosystem, groups, pages, hashtags, advertising tools, recommendation systems and reporting mechanisms, as well as strategies for evadingXThe campaigns posed as Americans on X and YouTube, used VPNs, and showed little evidence of achieving meaningful engagement.YouTubeThe campaigns posed as Americans on X and YouTube, used VPNs, and showed little evidence of achieving meaningful engagement.
Countries
ChinaOpenAI’s threat intelligence team identified two likely China-linked influence activity clusters that used ChatGPT to generate anti-data center and anti-tariff content aimed at inflaming U.S.United StatesOpenAI’s threat intelligence team identified two likely China-linked influence activity clusters that used ChatGPT to generate anti-data center and anti-tariff content aimed at inflaming U.S.