China's AI 'Poisoning' Poses Serious Threats, Risks of Information Warfare Emerge

Creating a realistic fake news story can take just a few hours, and it can quickly spread across the internet. Some netizens have noted that the propaganda department of the Chinese Communist Party (CCP) is the largest 'AI poisoning organisation' in the world. (People News/AI-generated image)

[People News] Chinese media recently uncovered a grey industrial chain involved in 'AI poisoning,' which has drawn significant attention. For just a few dozen yuan and a few hours of effort, nonexistent fake information can be presented as real by multiple mainstream AI models. Experts warn that this is not merely a case of commercial fraud; it could also evolve into a new form of information warfare in the AI era, exposing the fundamental flaws in large models' ability to verify the authenticity of information. Some netizens have pointed out that the CCP's propaganda agency is the largest forgery organisation in China.

AI Poisoning Forms an Industrial Chain

AI poisoning refers to the intentional creation of malicious data, misleading instructions, or contamination of training samples, which leads artificial intelligence models to produce false results, incorrect answers, or biased content, and even to output illegal or non-compliant information.

As reported by CCTV's '315 Evening Gala,' a reporter utilised a generative optimisation system to fabricate an 'Apollo-9' smart bracelet and automatically generated several promotional articles that included exaggerated and even absurd claims about its features. Within approximately two hours, multiple AI models were able to fully describe the product and even included it in their recommendation lists. By continuously disseminating false content, some mainstream models ranked it among their top recommendations.

According to a report from Phoenix Network, after publishing 11 pieces of misleading content over three consecutive days, several mainstream Chinese large models, including DeepSeek and Doubao, have ranked this fictional product among the top results for 'smart health bracelet recommendations'.

This industrial chain has established a complete commercial closed loop.

The exposed GEO (Generative Engine Optimisation) service charges an annual fee that ranges from several thousand to tens of thousands of yuan, and it can automatically generate and publish content in bulk.

The head of the GEO service provider revealed to the media that their company has served over 200 clients in just one year, spanning various industries such as healthcare, education, security, and renovation. They claimed, 'We can achieve a top-three ranking on any platform.' Another representative candidly mentioned that many major brands, in their quest for AI recommendation rankings, would consider 'spending millions to undermine the competition'; some businesses also use this tactic to launch smear campaigns against rivals.

The core logic, as stated by the service provider head, is: 'People don’t realise this is advertising; they will trust the results produced by AI.'

Media reports indicate that well-known Chinese large models such as DeepSeek, Doubao, Wenxin Yiyan, and Kimi fall within the client base of the exposed service provider.

Experts highlight that this phenomenon perpetuates the existing grey operation model of the Chinese internet, gradually evolving from search rankings to content marketing, and now to manipulation of AI. Since large models often depend on real-time retrieval to enhance timeliness but lack effective quality control over sources, this results in false information being more readily absorbed and amplified.

Analysis suggests that AI information manipulation has gone beyond traditional propaganda models. Previously, the focus was on influencing how users received information; now, it indirectly affects user decisions by manipulating AI judgments, making it hard for users to detect the underlying interference.

Experts stress that addressing these risks requires a comprehensive approach involving technology, regulations, and education to establish mechanisms for information verification and accountability. Additionally, enhancing public media literacy is crucial to preventing excessive trust in AI.

The public remains largely unaware that the answers they receive may be processed or even false, with a significant forgery organisation operating behind the scenes. (People News/AI-generated images)

The concept of AI information warfare is beginning to take shape.

Xie Peixue, an associate researcher at the Institute of Cybersecurity and Decision Simulation at Taiwan's National Defence Research Institute, told The Epoch Times that the application of AI in information manipulation has surpassed the realm of commercial fraud.

Xie Peixue explained, "The logic of traditional information warfare is 'to show you what I want you to see,' where users at least recognise that they are accessing information from various sources. In contrast, the information warfare of the AI era is about 'allowing AI to make judgments for you, and then manipulating those judgments,' often without users realising that the answers they receive are already influenced by commercial interests."

He noted that AI large models are fundamentally tools, and when data is systematically manipulated, these tools become distorted. The more profound issue is that 'human society is granting AI an authority it does not inherently possess,' with various forces—be it business, politics, or ideology—exploiting this false authority to further their own interests.

A reader of the People's Daily, known as 'Qin Yue', once commented: 'In China, the largest institution of deception is the propaganda machine controlled by the Chinese Communist Party. It portrays the Communist Party, which seized power through mass killings and brought immense suffering to the Chinese nation, as the 'great saviour of the people', and depicts dictators as 'great leaders'. During the three years of the COVID-19 pandemic, the extremely harsh zero-COVID policy led to a complete shutdown of the Chinese economy, resulting in the tragic disappearance of hundreds of millions of people. In light of this, many Chinese citizens have gradually awakened to the reality. They have come to a conclusion: 'Trust the Communist Party, and you will run straight into the crematorium.'”△