OpenAI has dismantled two covert, China-linked influence campaigns that leveraged ChatGPT to engineer fake social media personas and draft targeted propaganda. According to a threat report released Thursday, operators used the AI model to develop a detailed Facebook playbook designed to build credible identities, amplify messaging, and evade platform detection.
The first operation, likely orchestrated by a private Chinese technology firm contracting for provincial government clients, focused on American data centers. It generated content falsely claiming that AI infrastructure development was driving up residential electricity costs. The second campaign targeted US trade policy, producing anti-tariff messaging and political cartoons that exclusively featured President Donald Trump while deliberately omitting Chinese leadership.
Beyond content generation, operators prompted ChatGPT to design an automated surveillance system for monitoring online public opinion. OpenAI’s model provided only generic data management guidance and refused to assist with data collection tactics. Both networks utilized virtual private networks and simplified Chinese prompts to access the platform.
Despite the sophisticated operational security, OpenAI reports that neither campaign gained meaningful public traction. The company noted the strategic irony of Chinese actors relying on American artificial intelligence to critique US technology policy, underscoring a persistent pattern of foreign entities attempting to weaponize Western AI tools amid intensifying geopolitical competition.