Mechanism
How a Flag Becomes a Grassroots Task: The Departmental Chain Behind IJOP
Traces IJOP from government direction and data intake to platform flags, grassroots checks, and human disposition, while marking the limits of public evidence.
Contents
From an IJOP Flag to Grassroots Action
Public records confirm only parts of the chain; they do not reveal the full backend algorithm or every individual decision.
Xinjiang Trained Personnel on IJOP in 2016
On December 25, 2016, a Xinjiang regional vice chair hosted a training meeting for an integrated joint operations platform supported by big data. The government gazette did not publish the system's risk calculations, but it placed IJOP inside work organized at the autonomous-region level. [1]
Corporate records add the construction stage. Leon Technology's 2017 board report said that the company conducted surveillance projects in Kashgar and Aksu and undertook construction of Kashgar's integrated joint operations platform. [2] Regional officials organized training, local authorities purchased technical systems, and contractors built them. IJOP was never an algorithmic product operating apart from government. It was part of a Party-state security project.
Industry Data and Grassroots Governance Entered the Same Platform
A 2019 Xinjiang commerce document said the used-car trading system shared information with a political-legal commission platform and IJOP. Information about people, identity documents, vehicles, and images was compared in real time and uploaded to public security traffic authorities. [4] The document reveals one data entrance and shows how an ordinary commercial transaction entered a security system.
At the other end, IJOP connected to communities, police service stations, and grid management. A Sayibak District plan placed community IJOP work, convenience police stations, and grid management inside a normalized counterterrorism and stability system linked to rapid response. [3] Industry agencies supplied data, the platform formed leads, grassroots staff investigated, and police or other bodies acted. The algorithm had coercive power because an administrative chain stood behind it.
The Mobile Application Turned Flags into Investigation Tasks
Human Rights Watch reverse engineered a 2017 IJOP mobile application and found that workers could collect personal information, report people and vehicles, search by name, identity number, or address, and receive investigation tasks before returning results. Task fields concerned people, devices, locations, and relationships. [5]
The researchers had no account and did not log into the server to obtain live data. The mobile code therefore did not disclose the complete backend algorithm, actual lists, or every permission. It did reveal the grassroots operation. The platform organized data into flags and sent them to human investigators. A machine did not make every final decision alone, but people worked inside the field of suspicion the machine created. [5]
Internal 2017 Bulletins Connected Flags to Detention
Internal bulletins from 2017 published by ICIJ described what happened after IJOP generated lists. Workers investigated flagged people and then detained them or removed them from the list. Bulletin No. 14 stated that 24,412 people were marked suspicious in one week, with 15,683 sent to vocational education and training centers and another 706 formally arrested. [6]
The figures came from leaked documents examined by experts, not public judicial statistics, and they do not supply a complete record for each individual. They still expose the decisive institutional connection. Platform flags entered investigation, classification, and detention. The digital system widened the state's ability to locate risk populations, while grassroots agencies turned data judgments into physical coercion.
The 2022 OHCHR assessment similarly stated that public material appeared to show IJOP aggregating diverse data and flagging people based on conduct and indicators, with possible police follow-up and referral to training centers or other detention facilities. It examined the system within a differentiated surveillance environment directed at Uyghurs and other predominantly Muslim groups. [7]
A Software Error Can Harm Real People after Entering the Chain
An Associated Press investigation published in 2025 cited leaked emails, internal documents, procurement records, and interviews. It reported that engineers had fixed a software problem that incorrectly classified hundreds of people as high risk. [8] This was not a harmless defect confined to a test environment. Once a flag entered grassroots work, it could trigger investigation and disposition.
Responsibility cannot be assigned to the algorithm alone. Contractors built and maintained the system. Political-legal and public security organs organized data and tasks. Industry agencies supplied records. Community, police-station, and grid personnel conducted checks. Authorized bodies imposed the final measures. Software expanded the list, and the human institution gave the error coercive consequences. Each participant handled one part, while the combined machine offered the target almost no explanation. [2][3]
The Chinese Government Defended the Policy as Counterterrorism and Education
A 2019 Chinese government white paper described vocational education and training as lawful counterterrorism and deradicalization, emphasizing vocational skills and the national common language. [9] After the UN assessment appeared, the Foreign Ministry rejected it at a September 21, 2022 press conference as illegal, invalid, and assembled from false information. [10]
Those responses state the government's defense, but they did not disclose IJOP's flagging rules, error rate, individual lists, or appeal outcomes. The Party-state demands trust that the system prevents danger while denying flagged people a transparent route to inspect, correct, and challenge its records. National security expands the government's claim to know everything and removes the individual's right to know why punishment arrived.
The Public Record Has Limits on Backend Rules and Remedy
Existing records connect training, construction, data entrances, platform tasks, grassroots checks, and disposition. They do not publish the weight of each indicator, the method for generating complete lists, or permissions at every account level. Human participation in final disposition does not give a flagged person a formal route to inspect reasons, correct errors, and appeal. The software error reported by the Associated Press makes that gap concrete, but public material still cannot establish an overall error rate or show that every flagged person received the same treatment. [5][8]
IJOP Turns People into Risk Objects before Trial
IJOP's danger lies in the combination of machine and grassroots organization. The platform screens at scale, community and police personnel investigate, and government bodies decide what happens. Automation expands reach, the administrative system supplies force, and political labels make the process appear legitimate.
The CCP presents this arrangement as modern governance. In practice, it starts with suspicion, investigates the person, and makes the person bear the consequences. Algorithms are secret, lists are opaque, appeals are hard to see, and responsibility disperses across contractors and agencies. Technology did not remove human political judgment. It made the CCP's differentiated surveillance of Uyghurs and other Muslim groups faster, wider, and harder for its victims to challenge.
Sources: Xinjiang government gazette[1], Leon Technology annual report[2], Sayibak District plan[3], Xinjiang commerce material[4], Human Rights Watch reverse engineering of IJOP[5], ICIJ China Cables[6], OHCHR Xinjiang assessment[7], Associated Press investigation[8], Xinjiang vocational education and training white paper[9], and Foreign Ministry response[10].
What the available sources establish
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
Xinjiang Government Gazette, 2017 Issue 1primary-recordLive
Leon Technology 2017 Board Work Reportprimary-recordLive
Sayibak District 14th Five-Year Plan and 2035 Objectivesprimary-recordLive
Measures for Orderly Development of Xinjiang's Automobile Sectorprimary-recordLive
China's Algorithms of Repression: Reverse Engineering a Xinjiang Police Apptechnical-researchLive
China Cables: Operating Manuals for Mass Internment and Arrest by Algorithminvestigative-reportingLive
OHCHR Assessment of Human Rights Concerns in Xinjiangofficial-findingLive
How Silicon Valley Enabled China's Digital Police Stateinvestigative-reportingLive
White Paper on Vocational Education and Training in Xinjiangprimary-recordUnavailable
Foreign Ministry Spokesperson Wang Wenbin's Regular Press Conference on September 21, 2022primary-recordLive