Is the Chinese government misusing OpenAI's ChatGPT?
What's the story
OpenAI has accused the Chinese government of leveraging its AI chatbot, ChatGPT, for cyber operations. The allegation is detailed in a report titled "Disrupting Malicious Uses of AI." The document highlights how Beijing's operatives used the tool to refine internal status reports related to influence campaigns. The claim was first reported by Business Insider.
Tactics
Misuse discovered after account suspension
OpenAI discovered the alleged misuse after suspending an account associated with the Chinese government, which had been submitting periodic reports for ChatGPT editing. The company then launched a detailed investigation into China's "cyber special operations," aimed at potential dissidents within and outside China. These efforts were described as "large-scale, resource-intensive and sustained," involving hundreds of staff members, thousands of fake accounts, and dozens of tactics.
Targeted individuals
Targets included foreign leaders and global dissidents
The targets of these operations included global dissidents and even foreign leaders such as Japan's Prime Minister. The report detailed instances where Chinese operatives forged US court documents to pressure social media platforms into removing posts. OpenAI found that coordinated campaigns involved fake accounts submitting abusive reports in bulk to trigger bans on dissenting voices, with some even using AI-generated images as conversation screenshots.
Warning
Chinese Communist Party targeted popular X account
One high-profile target was the X account @whyyoutouzhele, aka "Teacher Li is not your teacher." The account has over 2.1 million followers and regularly shares videos exposing the corruption and human rights abuses in China. Its team warned that the Chinese Communist Party was weaponizing their content moderation system. They called on the AI industry to take responsibility for the technology being used to systematically oppress human rights.
Industry response
Other platforms have noted similar misuse patterns
Other platforms have also acknowledged similar patterns of misuse. Bluesky recently suspended accounts involved in a coordinated inauthentic activity, while Meta confirmed monitoring such operations in its adversarial reports. X and China's Ministry of Foreign Affairs did not respond to requests for comment on OpenAI's allegations.