OpenAI announced it banned accounts associated with a previously undisclosed China-origin influence operation, which it named "Nine-emdash Line." This operation utilized AI to produce regional influence content concerning the South China Sea, Hong Kong, and US political issues. The case study for "Nine-emdash Line" was initially published in OpenAI's October 2025 report.
The banned ChatGPT accounts were part of a covert influence operation originating from China, generating content for cross-platform distribution. Most of the content was in English, focusing on Vietnam's alleged environmental impact in the South China Sea and criticism of Philippines President Ferdinand Marcos. Cantonese-language posts targeted political figures and activists involved in Hong Kong's pro-democracy movement. A smaller portion of the operation generated English-language content on US political issues.
Key Points
- The operation, named "Nine-emdash Line," used ChatGPT accounts to generate influence content.
- Content focused on the South China Sea, Hong Kong's pro-democracy movement, and US politics.
- Posts were primarily in English and Cantonese, distributed on platforms like X and Instagram.
- The operation also used OpenAI's models for research, reconnaissance, and social media growth strategies.
- Some content resembled activity from the "Spamouflage" operation, which OpenAI discussed in May 2024, though no technical links were identified.
- The operation's impact was assessed as Category 2 on the IO impact Breakout Scale: activity on multiple platforms with minimal engagement.
Context
According to OpenAI, the social media accounts distributing the content bore some resemblance to the long-running China-origin operation known as "Spamouflage." However, OpenAI did not identify technical links between these operations. The Australian Strategic Policy Institute had previously identified some videos shared in the posts, linking them to an operation with similarities to earlier Spamouflage activity. The focus of this network appeared to be bulk generating social media posts involving countries with territorial disputes with China in the South China Sea, a region China claims via the 'Nine-Dash Line.' The operation's use of AI-generated text that included em-dashes in its posts contributed to its name, "Nine-emdash Line."
Why It Matters
This disclosure highlights how AI models are being used in state-linked influence operations, requiring builders and practitioners to consider the potential for misuse in content generation, research, and strategic planning. The low engagement assessment suggests that while the capability exists, the immediate impact of such operations can vary.
What To Do
- Note the specific content themes and languages targeted by the "Nine-emdash Line" operation.
- Observe how AI models were used for both content generation and auxiliary tasks like research and social media strategy.
- Review the OpenAI October 2025 report for further details on this case study.
- Watch for future reports from OpenAI on malicious uses of AI, such as the February 2026 report detailing "Trolling Stone."