When an official news program on state-run Hunan Economic Television took to the air on May 6, the tension between artificiality and authenticity was on full display. As two AI-generated anchors named Sheng Sheng and Shuang Shuang read out the news, a message appeared in the corner of the screen: “content not [AI] generated.” The disclaimer was an about-face from the broadcast two days earlier, on May 4, when the AI-generated anchors had debuted with a simple “AI generated” label that unleashed a torrent of public criticism.
Online commentators were outraged at the thought that the news content, too, had been crunched with AI. Were human journalists being replaced entirely? The apparent misunderstanding prompted the broadcaster to add new language, clarifying that while the anchors’ images were AI-generated, all reporting, writing, and editing remained in human hands.
The episode, analyzed by the China Media Project on May 15, exposes a tension running through China’s state media landscape, one that is unlikely to be resolved by a disclaimer in the corner of the screen.
Propaganda Controls Remain Intact
An important caveat is absent from most coverage of the incident: those human hands are under strict propaganda controls from Party-state authorities in Hunan. According to registry documents, Hunan Economic Television is a subsidiary of Hunan Radio, Film and Television Group, which is wholly controlled by the Propaganda Office of the Hunan Provincial Committee of the CCP.
The China Media Project notes that the public’s reaction reflects a quiet anxiety that goes deeper than the question of who writes the scripts. A news culture already distrusted for its proximity to power may now outsource even the appearance of human judgment to machines, and human journalists, whether or not they are doing real journalism, may pay the price. The backlash was not primarily about AI safety or accuracy; it was about the perceived erosion of the last visible sign that a human being was responsible for what appeared on screen.
Chinese authorities have enthusiastically promoted AI as a tool of media modernization. China’s state media has been deploying AI-generated content since at least 2024, and the trend has accelerated as the technology has improved. But the Hunan case illustrates that the rollout is not frictionless. Viewers who have grown accustomed to synthetic faces in entertainment contexts, a market where iQiyi has enrolled more than 100 AI actors, appear to draw a sharper line when the same technology is applied to news.
The Hangzhou Court Ruling as Context
The China Media Project connects the Hunan backlash to a separate but related story that has been circulating in Chinese media this month: a court in Hangzhou recently ruled that a financial technology firm had acted illegally when it demoted and ultimately dismissed a 35-year-old AI quality-control supervisor on the grounds that his work could be performed by AI. The court ordered the company to pay over 260,000 yuan in compensation.
A commentary published by Dahe Online, a news portal overseen by the Propaganda Office of the Henan Provincial Committee of the CCP, said the ruling “punctures the myth of AI omnipotence.” The juxtaposition is striking: a state-affiliated media outlet using a court ruling about AI-driven job loss to push back against the idea that AI can replace human workers, even as state-affiliated television stations are doing exactly that with their news anchors.
The contradiction points to a genuine policy tension in Beijing’s AI strategy. The central government has been aggressive in promoting AI adoption across the economy, including in media. But it has also been sensitive to public anxieties about job displacement, particularly in the context of a labor market that has been under pressure since the pandemic. The Hangzhou ruling and the Hunan backlash are both symptoms of the same underlying unease.
The Broader Stakes for AI in Chinese Media
The Hunan incident is not an isolated case. China’s television industry has been experimenting with AI anchors since at least 2018, when Xinhua News Agency unveiled what it described as the world’s first AI news anchor. The technology has improved dramatically since then, and the cost of deploying synthetic anchors has fallen to the point where provincial broadcasters can afford to experiment.
But the public reaction in Hunan suggests that the technology’s improving quality has not resolved the fundamental question of legitimacy. Viewers do not just want to know whether the anchor is real; they want to know whether the institution behind the anchor is accountable. A disclaimer saying “content not AI generated” answers the first question but not the second, and in a media environment where the content is shaped by Party propaganda directives regardless of whether a human or a machine assembles it, the distinction may matter less than the broadcaster hoped.
The episode also raises questions about how AI adoption in Chinese state media will interact with the country’s broader AI governance agenda. China has issued regulations requiring AI-generated content to be labeled, and the Hunan broadcaster was in compliance with those rules when it added the “AI generated” label on May 4. The backlash suggests that compliance with labeling requirements is not the same as managing public trust, a lesson that applies well beyond the state media context.