Mechanism
From Fragmented Records to Police Leads
Traces how police platforms ingest, standardize, and link records into leads or check tasks, while separating capacity, deployment, operation, and consequences.
Contents
Police Data Fusion Chain
This observable chain combines public design documents with IJOP technical evidence; it does not show that every locality deployed and operated every stage.
Five Evidentiary Levels
Evidence at an earlier level cannot prove later operation or consequences.
| Evidence Level | Can Establish | Cannot Establish |
|---|---|---|
| Stated Goals | Construction and operational requirements | Actual procurement or deployment |
| Procured Capacity | Functions listed in procurement | That functions entered operation |
| Local Deployment | Projects publicly described as deployed | Use of a function in a specific case |
| Technical Operation | Observable interfaces and operation | All back-end algorithms and permissions |
| Case Consequences | Recorded tasks, checks, or actions | The same consequences in other regions |
How fragmented records become police tasks
Police big-data projects bring records once separated across household registration, cases, vehicles, addresses, and organizations into a shared environment. They can add information from other government departments, companies, internet sources, and video systems. Public technical material describes ingestion, cleaning, field standardization, entity linkage, and delivery to operational systems [1]. What an officer finally sees may be a personal file, a relationship map, an alert, or a task ordering an in-person check.
National identification numbers, phone numbers, and addresses can link records from different databases to one person. Shared organizations, contacts, vehicles, and cases extend the file into a network. An investigation of early police-cloud procurement documents in Shandong, Jiangsu, and Tianjin found plans to connect government and corporate data for relationship discovery and anomaly detection [9]. Errors travel through the same chain. A wrong code, obsolete address, or broad label can become a fresh risk signal in several operational systems.
Local platforms keep expanding
Jiangsu described a province-and-city smart-policing structure in 2018. Shandong described links among police-branch records, social data, and front-line applications in 2019 [2] [3]. A 2023 Hunan police response described a provincial cloud alongside city and prefecture investment. A 2025 Anshan award divided resource leasing, data aggregation, intelligent applications, and a new integrated police platform into procurement packages [4] [5].
These records show continuing procurement and deployment, but they do not prove that every city uses the same database, algorithm, or access rules. A design document establishes an objective. A tender establishes purchased scope. A local publicity article establishes what officials claim to have deployed. Operational records and case evidence are needed to show how a function affected a person.
Xinjiang exposed an operational loop
The mobile application for Xinjiang's Integrated Joint Operations Platform offered evidence closer to actual operation. Human Rights Watch's reverse engineering identified interfaces for person searches, files, information collection, and task delivery [10]. Researchers did not log into the server and could not reconstruct every database, algorithm, or label. The client structure nonetheless showed how aggregation, risk flags, and checks on people could form a loop.
The UN human rights office's 2022 assessment found that available information indicated extensive aggregation of personal data and the ability to flag people as potential detention candidates [11]. China's Foreign Ministry rejected the assessment's legitimacy and credibility [14]. That general denial did not explain each interface identified in the client or disclose false-positive rates, correction records, and appeal channels.
From technical construction to political selection
The Personal Information Protection Law requires state bodies to process personal data within statutory authority and necessity. The Data Security Law requires police data access to follow rules and approval procedures [7] [8]. Those formal provisions do not replace scrutiny of actual interfaces, query logs, and task rationales. A flagged person usually cannot learn which databases produced the judgment or demand removal of a false link.
CCP publicity about smart policing stresses speed, prediction, and coordination while revealing little about who defines risk, who can cancel an erroneous task, or how political labels enter a model. The danger is not that the machine is more intelligent than an officer. It is that the party-state can convert scattered surveillance into lists executable at scale. A political suspicion that once required a chain of cadre instructions can now arrive directly on a local officer's device.
Disputes and evidence limits
The public record firmly supports four points: police agencies have built cross-database aggregation capacity, local projects include standardization and relationship analysis, some systems deliver results to front-line personnel, and technical material from Xinjiang exposes interfaces between tasks and checks on people. It is reasonable to infer that data quality, linkage rules, and risk labels affect who is selected for scrutiny. The evidence does not establish that every police platform in China uses one algorithm, every procured feature is operating, every alert ends in detention, or every offline action is triggered directly by an automated model [1] [10].
Official project descriptions celebrate more data, faster linkage, and more precise alerts. They rarely disclose false-positive figures, label expiration rules, access-control changes, or outside audits. Procurement records do not reveal which delivered modules passed acceptance, which rules were later disabled, or which office can view a complete relationship network. Treating planned capacity as proven operation would conceal the decisions that deserve scrutiny.
China's rejection of the UN assessment must be recorded accurately as the government's position, but a general denial does not answer each technical finding [14]. Outside investigations also retain geographic, temporal, and methodological limits. Those limits do not weaken the central judgment. The CCP refuses to expose interfaces, logs, and correction data while agencies controlling the information also control coercion and access to remedies. A flagged person has no comparable ability to inspect or correct the system.
Sources
- [1] State Information Center paper on police data-analysis platforms.
- [2] Official account of Jiangsu smart policing.
- [3] Official account of Shandong police informatization.
- [4] Hunan Provincial Public Security Department response.
- [5] Anshan police big-data procurement award.
- [9] Human Rights Watch investigation of police clouds.
- [10] Human Rights Watch reverse engineering of the IJOP app.
- [11] UN human rights office assessment of Xinjiang.
- [7] Personal Information Protection Law of the PRC.
- [8] Data Security Law of the PRC.
What the available sources establish
A 2023 public reply from the Hunan Provincial Public Security Department stated that procurement for five first-phase provincial public-security big-data infrastructure projects had been completed with investment exceeding RMB 300 million, while projects by 14 prefecture-level public security authorities exceeded RMB 650 million.
Technical analysis of an application linked to Xinjiang's Integrated Joint Operations Platform found that it aggregated multiple forms of personal data and generated leads for police follow-up.
A 2025 Associated Press investigation reported, based on leaked materials, that hundreds of thousands of people had been tagged as ‘untrustworthy,’ that IJOP flagged 24,412 people as ‘suspicious’ in one week in 2017, and that engineers fixed a software bug to release hundreds of people categorized as high risk.
Sources
Building a Public Security Data Analysis Platform in the Big Data Eratechnical-researchLive
Jiangsu Police Build a New Smart Policing Systemprimary-recordLive
Shandong Localities Deepen Public Security Informatizationprimary-recordLive
Reply to Recommendation No. 1312 of the First Session of the 14th Hunan Provincial People's Congressprimary-recordLive
Award Notice for the Public Security Big Data and New Integrated Policing Platform Projectprimary-recordLive
Ministry of Public Security KXX2205 Project Public Tender Noticeprimary-recordLive
Personal Information Protection Law of the PRCprimary-recordLive
Data Security Law of the PRCprimary-recordLive
China: Police ‘Big Data’ Systems Violate Privacy, Target Dissenthuman-rights-reportLive
China's Algorithms of Repression: Reverse Engineering a Xinjiang Police Apptechnical-researchLive
OHCHR Assessment of Human Rights Concerns in Xinjiangofficial-findingLive
How Silicon Valley Enabled China's Digital Police Stateinvestigative-reportingLive
Regulation on Public Security Video Image Information Systemsprimary-recordLive
Foreign Ministry Spokesperson Wang Wenbin's Regular Press Conference on September 21, 2022primary-recordLive