State actors lean on AI narratives
China’s state media is increasingly using social media and AI-generated content to shape narratives and to mock foreign rivals, showing that synthetic media and AI-native persuasion are becoming standard tools in institutional communication. That trend means provenance, trust signals and community reputation are likely to matter more for any product that relies on public-facing content or discovery. (yoursourceone.com)
China’s state broadcaster is now posting artificial-intelligence cartoons in which the United States appears as a white eagle, Iran appears as a band of Persian cats, and a five-minute martial-arts allegory travels on social media instead of on the evening news. The shift is not that Beijing discovered propaganda in 2026; it is that state messaging now looks like internet-native entertainment built for feeds, reposts, and younger viewers. (apnews.com) The same playbook showed up a year earlier in the tariff fight. On April 3, 2025, China Global Television Network, the state broadcaster’s international arm, published a 2-minute, 42-second artificial-intelligence song aimed at Donald Trump’s tariffs, and Xinhua’s New China TV published a 3-minute, 18-second sci-fi short called “T.A.R.I.F.F.” the same day. (time.com) That matters because the format changed along with the message. A foreign ministry statement asks for patience and attention; a catchy song, a meme, or a cartoon asks for a share, and platforms like YouTube, X, TikTok-style feeds, and repost chains reward material that is fast, visual, and emotionally simple. (time.com) China has spent years learning how to work those systems. In April 2024, Microsoft said China-linked influence accounts were using fake social media profiles to probe divisive issues in the United States and had increased their use of artificial-intelligence content on topics including immigration, racial tension, drug use, the Maui wildfires, and Japan’s wastewater release. (blogs.microsoft.com) Researchers describe this as a move from clumsy bot spam to something closer to mass-produced persuasion. RAND wrote in December 2024 that artificial intelligence makes the “messenger” more plausible, blurs the line between real and fabricated material, and lowers the cost of creating the illusion that lots of ordinary people all believe the same thing. (rand.org) The striking part is that Beijing is pushing synthetic media outward while tightening rules at home. In September 2025, Chinese platforms including WeChat, Douyin, Weibo, and RedNote began rolling out labels for artificial-intelligence-generated text, images, audio, and video under a new law backed by the Cyberspace Administration of China and three other agencies. (engadget.com) So the same state that wants labels, metadata, and platform enforcement inside China is also showing how useful unlabeled-looking style and platform fluency can be when the audience is abroad. That is the real change: state communication is starting to look less like a press release and more like a creator account with a geopolitical brief. (engadget.com) (apnews.com) For anyone building search, discovery, news, moderation, or creator tools, the pressure point is no longer just whether a post is false. The harder question is whether a polished video, song, or meme came from a person, a model, a government media desk, or all three stitched together in one workflow. (rand.org) (engadget.com) That is why provenance systems, source labels, and reputation signals are moving from nice extras to basic plumbing. When a state outlet can publish a cartoon that looks like internet culture and an influence network can test which arguments spread best, the old shortcut of “it feels authentic” stops being much of a shortcut at all. (rand.org) (blogs.microsoft.com)
