Guerrilla influencers and AI news anchors, China is ramping up its propaganda machine

From influencers to state media, China’s propaganda machine is embracing AI – but there are limits to what tech can do with bad storytelling, writes Ruby Osman

The generative AI game has a new player: Chinese state media. Over the past month, CGTN, the international division of China Central Television, has been unveiling a slick new AI-generated video series. Fractured America is a series of glossy video shorts that take aim at class divides, labour disputes and the military-industrial complex through the eyes of (cartoonified) ordinary Americans.

It’s uneven in places, outright bizarre in others and the voiceover leaves a lot to be desired – and yet, it’s immediately head and shoulders above the usual state media output.

The generative AI boom has come at just the right time for China’s English-language state media. It’s been more than a decade since Xi Jinping first instructed officials to “tell China’s story well”, but as Beijing eyes new markets in the Global South and feels the US is turning increasingly hostile, journalists are coming under renewed pressure to get China’s message across in the global information space. It’s not enough just to be seen as an economic partner – under Xi, the brief is to seek rightful recognition of China’s status as a cultural, geopolitical and historical power.

Part of that push is, of course, highlighting China’s achievements in domestic development and its role as a global force for stability. But just as important is contrasting that with the US – a superpower that state media frames as dangerously divided at home and dangerously demanding abroad.

The result is content like Fractured America, which makes the case that defence companies are “turning global conflicts into cash” and that “politicians talk sweet, but then high-five the high rollers”.

It’s a multi-media effort too: Xinhua, China’s official news agency, earlier this month took aim at the US’s debt levels in a new AI-generated song, shared on X, that claims: “Though debts are scary and solutions unfound, with dollars in hand we can dance around.” The message is driven home by a music video that features a series of increasingly grotesque cartoons of Uncle Sam. As with all the best propaganda, it works because it draws on the genuine concerns of many Americans.

And that ability to needle away at the US’s internal divisions is only likely to get better – this is just the start of a learning curve. The propaganda potential of generative AI is immense: think better visuals, more natural language, fewer resourcing demands and, most crucially, smarter and more sophisticated ways to figure out what it is that actually cuts through to foreign audiences. Videos, text and images are all likely to get sleeker and more effective as media outlets combine new technologies with old grievances.

And for all the fear of Tiktok, this is a toolbox that will work on any platform. AI content can – and already has – spread across Youtube, Facebook and X. It’s not just state media that are picking up on calls from the top to “tell China’s story well” either: a wave of “guerrilla influencers” are starting to push out their own AI-generated content.

In December last year, the Australian Strategic Policy Institute identified what it claims is a “coordinated inauthentic influence campaign originating on Youtube”, with a network of at least 30 Youtube channels producing more than 4,500 pro-China and anti-US videos. The network appears to be run by a commercial, not state, actor, although the authors note some level of state direction could be possible.

And in February last year, research firm Graphika identified two videos from “deepfake” news anchors reporting for fictitious outlet Wolf News, including one that took aim at the US’s lack of action against gun violence. Again, it’s unclear exactly who produced the videos, but the inspiration may well have come right from the top: China’s official state news agency Xinhua unveiled the world’s first AI news anchors all the way back in 2018.

But, however sophisticated the tech gets, China’s state media outlets will still struggle to overcome the paradox at the heart of “telling China’s story well”: the need to let go of control to gain control.

It’s easy to bash the US, but it’s proving harder to win foreign audiences over to China. That’s in part because Chinese outlets instinctively take a “constructive” approach to reporting on China, focusing on the good wherever possible and taking a positive, solutions-focused approach to tackling social problems.

This is par for the course within China – in fact, the term “propaganda” has little pejorative meaning in Chinese – but it’s less convincing abroad, where relentlessly positive coverage of China sits uneasily among audiences used to their media eco-systems criticising, not championing, government.

That means that any attempt to better control the global information space will, against all the natural impulses of a tightly regulated system, likely require a willingness to cede some control. There’s a ceiling to propaganda efforts as long as “telling China’s story well” is interpreted as “only talking about China’s good side”.

Some of China’s top intellectual voices recognise these limitations. As Zichen Wang points out in his Pekingology newsletter, former director of the State Council Information Office Zhao Qizheng made the case back in 2020 for telling the full story of China’s “70 per cent brightness and 30 per cent backwardness”, i.e. not just focusing on the “brightness”. Or as Henry Kissinger once put it more bluntly to Chinese officials: “Who will believe that there is a country in the world with no shortcomings?”

Jia Qingguo, director of Peking University’s Institute for Global Cooperation and Understanding, made a similar case during this year’s Two Sessions. In his view, a plurality of voices from China would help drive home China’s story. Instead, “excessive restrictions” on who can travel abroad and speak to foreign media means that China’s official output is disregarded as “mere propaganda”.

Taking that advice on board, however, is easier said than done. Writing in a way that’s attuned to the distinct needs of foreign audiences is a big ask in a system that increasingly incentivises “writing for an audience of one” – in other words, putting out what journalists think top leadership wants to see.

But without adapting, China will find there are limits on what deep-fake news anchors, glossy AI graphics and punchy headlines can do. Criticising the US is one thing, but making a compelling case for China might just prove harder.