In a concerning revelation, Microsoft has uncovered a network of social media accounts controlled by China, which employs artificial intelligence (A.I.) technology to impersonate American voters and disseminate propaganda with the aim of influencing U.S. politics.
Microsoft’s recent report reveals that this suspected Chinese influence network bears resemblance to past operations that have been traced back to the Chinese Ministry of Public Security. However, what sets this new network apart is its use of generative A.I. since March 2023. Unlike previous campaigns, this network employs A.I. to “mimic U.S. voters” and create content that is far more eye-catching than the less sophisticated visuals used in previous Chinese state-sponsored campaigns.
Microsoft suggests that these compelling images are most likely generated using diffusion-powered image generators, which utilize A.I. not only to produce attractive visuals but also to continuously enhance their quality over time.
This Chinese influence network has launched a broad campaign that predominantly focuses on politically divisive subjects such as gun violence while disparaging U.S. political figures and symbols. To illustrate, Microsoft researchers presented an A.I.-generated image featuring the Statue of Liberty wielding an assault rifle beneath the caption “Everything is being thrown away. THE GODDESS OF VIOLENCE.”
Another example was a Black Lives Matter poster that seemed to be created by a “Chinese Communist Party-affiliated automated account” and was later uploaded by an account impersonating a U.S. conservative voter.
While these social media posts may not appear particularly persuasive, and their origin as Chinese propaganda may not be well-disguised, Microsoft’s concern lies in the minimal human intervention required to generate such content. The A.I. behind these posts is continually learning from its mistakes, enhancing its propaganda capabilities. Researchers also noted that these Chinese influence networks are becoming more adept at responding to human interactions.
According to Microsoft, the China-linked accounts attempted to appear American by listing their public location within the United States, sharing American political slogans, and using hashtags related to domestic political issues.
The report by Microsoft also highlighted that Chinese cyber activity continues to pose a threat to the United States. The top three subjects of Chinese hacking activity in 2023 include the South China Sea, the U.S. defense industrial base, and U.S. critical infrastructure.
This revelation underscores the growing use of A.I. in the realm of disinformation campaigns. While the current attempts may appear unsophisticated, they serve as a warning sign of the potential for more advanced and convincing A.I.-generated propaganda in the future. As nations grapple with the implications of A.I. in the information war, it becomes increasingly important to develop strategies to combat and detect such threats.
