China's AI Police Technology Is Watching — and Feeling — Everything
At a sprawling law enforcement equipment exhibition held in Beijing, Chinese technology companies pulled back the curtain on a new generation of AI-powered police tools that go far beyond facial recognition. These systems, demonstrated to domestic and international attendees, are designed to assess the physical health, mental state, and even the emotional risk profile of individuals in police custody or under surveillance. The implications for law enforcement efficiency are significant — and the implications for civil liberties are even more so.
What Was on Display at the Beijing Law Enforcement Expo
The international police and anti-terrorism technology expo, which ran for three days in Beijing, featured dozens of Chinese firms showcasing their latest biometric devices and AI-integrated systems. Exhibitors emphasized that their products could help reduce the manpower burden on a police force that is reportedly stretched thin, improving operational efficiency amid a recognized shortage of frontline officers across China.
The demonstrations covered a wide range of capabilities, but the most striking were those focused on reading and interpreting human physiological and psychological signals. Rather than simply identifying who a person is, these systems are built to determine what a person is feeling, thinking, and potentially planning.
Assessing Physical Health: Beyond the Basics
Some of the devices on show were designed to rapidly evaluate the physical condition of detained suspects. In a traditional policing context, assessing whether a detainee is injured, unwell, or under the influence of substances typically requires trained medical personnel or significant time. The AI-powered tools demonstrated in Beijing aim to automate this process, using sensors and cameras to detect signs of physical distress, intoxication, or illness within seconds of contact.
This capability, proponents argue, can reduce liability risks for police departments and ensure that suspects receive timely medical attention. Critics, however, point out that automating such assessments removes human judgment from a process that has serious legal and ethical dimensions — and introduces the risk of algorithmic error at potentially life-altering moments.
Psychological and Emotional State Detection: The Frontier of AI Surveillance
Perhaps the most controversial element of what was on display is the push to detect psychological and emotional states. Several exhibitors claimed that their systems could evaluate a subject's mental condition — identifying signs of anxiety, aggression, deception, or psychological instability — using a combination of facial analysis, voice pattern recognition, and physiological monitoring such as heart rate and galvanic skin response.
The idea of a machine being able to tell police whether a suspect is lying, agitated, or mentally unstable is the kind of technology that once belonged in science fiction. But the companies presenting at the Beijing expo described these systems as operational, deployable, and already in various stages of pilot testing.
The technology draws on decades of research into affective computing — the field dedicated to building machines that can recognize and respond to human emotions. While researchers in this space have made genuine progress, independent studies have repeatedly raised concerns about accuracy, cultural bias, and the fundamental difficulty of reliably inferring internal states from external signals.
Risk Profiling: Who Is Considered a Threat Before They Act?
Layered on top of the health and emotional assessments is the concept of AI-driven risk profiling. Some systems demonstrated at the expo were presented as capable of generating a risk score for an individual — an algorithmic judgment about how likely someone is to pose a threat or cause problems. This feeds directly into predictive policing frameworks, where the goal shifts from responding to crime to anticipating it.
This is where the technology becomes particularly contentious from a human rights perspective. Predictive risk systems have been challenged in multiple jurisdictions around the world for embedding existing biases into their models, effectively penalizing individuals based on factors over which they have no control. In the Chinese context, where minority communities such as Uyghurs have already been subjected to intensive biometric surveillance, the prospect of AI risk scoring raises acute concerns among international human rights observers.
Efficiency Gains vs. Ethical Costs: The Central Debate
Chinese authorities and the companies behind these tools frame the technology primarily as a solution to a practical problem: too many policing tasks, too few officers. By automating assessment, screening, and monitoring tasks, AI-enabled equipment can theoretically free up human officers to focus on more complex work. Exhibitors at the Beijing expo pointed to reduced processing times, lower rates of human error in routine screening, and better resource allocation as the headline benefits.
But the efficiency argument cannot be cleanly separated from questions about how these tools will be used, on whom, and with what oversight. Law enforcement technology that works reliably in controlled demonstrations does not always perform the same way in the messy reality of policing — and when it fails, the people most likely to bear the consequences are those who are already most vulnerable to over-policing.
A Global Technology Race With Local Rules
China is not the only country developing AI tools for law enforcement. Governments and police agencies across Europe, North America, and Asia are investing in biometric screening, behavioral analysis, and predictive systems. What sets the Chinese development landscape apart is the speed of deployment, the scale of the domestic market, and the relative absence of the legal and regulatory friction that slows — and sometimes stops — similar programs in democratic countries.
The Beijing expo also had an international dimension, with foreign buyers and observers present. The global export of Chinese police AI technology has been a growing concern for human rights organizations, who warn that these tools can be acquired by authoritarian governments with few safeguards for how they will be applied against their own populations.
What Comes Next
The technology demonstrated in Beijing represents a significant step forward in the integration of AI into the full policing process — from the moment a person is stopped or detained, through assessment, risk classification, and beyond. As these tools mature and become more widely deployed, the conversation around regulation, transparency, and accountability will become increasingly urgent — both within China and in every country where similar technologies are being developed or imported.
The question is no longer whether AI can read your body language, assess your emotional state, or predict your behavior. The question is who gets to decide when that capability is used, and who is held responsible when it gets it wrong.