OpenAI has banned accounts linked to the China-origin influence operation known as “Spamouflage.” This network, attributed by Meta to individuals associated with Chinese law enforcement and by the FBI to China’s Ministry of Public Security, targeted global audiences, particularly the Chinese diaspora and critics of the Chinese government.

The operation utilized OpenAI's models for tasks such as debugging code for the website revealscum.com, which published criticisms and personal information about diaspora members labeled as “traitors.” They also generated content accusing Japan of environmental damage from Fukushima wastewater and criticized Chinese dissident Cai Xia. The AI was used for open-source research, including social media account applications and summarizing posts by government critics.

Despite using AI, Spamouflage’s output appeared to be a minority of their total activity, with much content seemingly manually created. The operation’s impact remains minimal, with little evidence of authentic audience engagement across platforms like Medium and X. OpenAI assessed this as a Category 2 operation, characterized by widespread posting but no significant amplification by external individuals, with revealscum.com identified as a related domain.