From watching citizens to predicting behavior, China’s surveillance system is entering a darker and more dangerous phase

For years, the Chinese Communist Party has built what many observers consider the largest surveillance system in human history. Hundreds of millions of cameras, facial recognition tools, centralized databases, and digital tracking systems have spread across Chinese cities and towns, creating a society in which ordinary people can be watched almost anywhere they go. But according to recent reporting and government procurement records, the CCP is now pushing into something even more disturbing: a system designed not merely to observe people, but to predict them.

This shift represents a profound escalation in state control. China is moving from a reactive surveillance model, where authorities investigate wrongdoing after something happens, to what officials and analysts describe as “predictive policing,” an approach intended to identify risks before crimes, protests, or disruptions occur. In plain English, this means people may increasingly find themselves scrutinized, investigated, or targeted not for what they have done, but for what authorities believe they might do.

How China Got Here

The new findings come from reporting by the Financial Times, whose journalists reviewed more than a dozen Chinese government tenders and interviewed people familiar with surveillance contracts. Their conclusion was stark: local authorities throughout China are rapidly upgrading surveillance infrastructure with advanced artificial intelligence systems capable of analyzing behavior and identifying potential threats automatically.

China already possessed an enormous surveillance apparatus. Existing systems such as “Skynet” and “Sharp Eyes” helped blanket urban and rural regions with cameras and tracking systems. State media once boasted that Skynet could scan China’s population in under a second, while Sharp Eyes extended surveillance into smaller communities and, in some places, integrated private cameras into centralized systems.

The surveillance network had already proven useful for political control. As the reporting notes, China used previous versions of these systems to monitor the Uyghur population in Xinjiang, where surveillance became a tool for extraordinary state oversight. But experts say older systems had a major limitation: they were better at watching people who were already under suspicion than at identifying future threats.

Now, the CCP wants something much more ambitious.

The Rise of Predictive Policing

The defining feature of China’s modernization effort is the transition to predictive policing. Older systems required police officers to spend enormous amounts of time reviewing footage manually. The new systems are designed to work proactively, automatically flagging suspicious activities and warning authorities in real time.

The technology is powered by a combination of computer vision, large language models, and more advanced chips supplied by firms such as Hikvision and Huawei. Instead of simply storing video, cameras can increasingly analyze it as events unfold. In many cases, processing occurs directly on cameras or nearby servers rather than distant data centers, allowing faster decisions and lower costs.

Authorities say these systems can identify dangerous or abnormal behavior. The technology may flag sudden road maneuvers, unauthorized gatherings, illegal entry into restricted spaces, or even signs of suicidal intent when someone spends too much time standing near a bridge. Hikvision systems reportedly allow operators to search video with simple written descriptions such as “woman in a red hat,” eliminating hours of manual review. One company manager reportedly told the Financial Times that officers no longer need to spend hours searching footage because “it’s enough to enter a text query.”

That capability may sound efficient. But efficiency in the hands of an authoritarian government comes with consequences.

In democratic societies, surveillance is usually constrained by legal protections, judicial oversight, and the presumption that citizens are innocent until proven guilty. China’s evolving system points in another direction entirely. The logic underpinning predictive policing assumes the state should intervene before wrongdoing occurs, even if what is being predicted amounts to suspicion rather than evidence.

This is where the repression becomes deeply troubling. When governments begin evaluating citizens according to what they might someday do, ordinary freedoms become conditional. Suspicion replaces evidence. Algorithms replace due process.

A System Built to Control More Than Crime

Chinese authorities frame these developments as crime prevention. Yet the supplied reporting suggests something much broader is underway.

The technology is designed not only to identify criminal acts, but also to monitor public gatherings, emotional states, and possible “social governance incidents,” a government phrase associated with protests or public unrest. Researchers documented efforts to process personality traits, long term emotional states, and even exposure to what authorities call “negative cultural influences” in an effort to forecast instability before it happens.

That distinction matters enormously.

A camera that spots a robbery is one thing. A system that attempts to determine whether citizens are emotionally unstable, politically restless, or potentially troublesome is something very different. It crosses the line from law enforcement into social engineering.

One reported patent proposal from a university in Guizhou described using advanced AI reasoning models to predict so called social governance incidents before they emerge. Such systems depend on enormous quantities of personal data gathered through cameras, local reporting networks, and digital monitoring. Since early 2025, “grid workers” in neighborhoods have reportedly uploaded real time observations through smartphone applications, feeding local data into AI systems that analyze community sentiment and risk at extraordinary scale.

The result resembles a society in which every citizen becomes a subject of continuous evaluation.

Citizens are not merely watched. They are scored, categorized, and interpreted.

The implications are chilling. A government that can predict who may gather in protest, who appears emotionally distressed, who spends “too much” time in a certain location, or who seems influenced by politically disfavored ideas acquires immense power to intervene long before dissent materializes.

This is preemptive social control disguised as public safety.

Institutionalizing an AI Police State

Perhaps most alarming is that China’s surveillance ambitions are no longer experimental. They are increasingly formalized in law and government planning.

The reporting notes that China’s revised cybersecurity framework and national planning documents explicitly incorporate AI driven governance and cyber control into long term state strategy. Researchers also documented growing AI integration into criminal justice systems, including platforms that recommend prosecutorial actions, assist indictments, and even influence judicial proceedings. In some prisons, facial recognition systems reportedly monitor inmates’ emotional expressions, flagging individuals if they appear angry.

The danger here is not merely surveillance. It is automation.

A human police officer can be questioned. A judge can be challenged. Evidence can be examined.

But algorithmic suspicion is harder to confront because it is often invisible. Why was someone flagged? What data triggered concern? What emotional profile or behavioral category placed them on a watch list? Citizens may never know.

This creates the conditions for a society where punishment or scrutiny arrives before wrongdoing, justified by predictions that cannot easily be challenged.

The Chinese Communist Party has long sought obedience and social stability. Advanced AI gives it something even more powerful: the ability to identify friction before it erupts and suppress it before citizens even realize they are being targeted.

The Human Cost of Predictive Repression

The central tragedy of China’s surveillance transformation is that ordinary citizens increasingly risk becoming suspects in their own lives.

A man standing quietly near a bridge, a crowd gathering peacefully in a public square, a citizen experiencing emotional distress, or someone expressing dissatisfaction may all become inputs in an immense machine designed to classify risk. In such a system, behavior itself becomes evidence.

The CCP describes this as modernization and public safety. Critics see something much darker: the industrialization of political obedience.

A government willing to punish people because they might become a problem has crossed into profoundly dangerous territory. The cameras no longer merely record reality. They are increasingly being used to predict, categorize, and control it. And for the Chinese people, that may be the most frightening development of all.