Mechanism

Skynet and Sharp Eyes: How Central Targets Become County Procurement

How central targets cascade from nine agencies through the comprehensive management system, local budgets, and county procurement in China's surveillance infrastructure.

Core Question

In narratives about Chinese surveillance, "Skynet" is often presented as a single nationwide system. Organizationally, it functions more as two parallel infrastructure lines: within cities, the Skynet and Safe City programs led by public security bureaus, and extending into county, township, and village levels from 2015 onward, the Sharp Eyes Program (xueliang gongcheng, 雪亮工程) directed by the central Party's Political and Legal Affairs Commission through its comprehensive social management system ^[11][8].

Understanding how camera networks blanket a county is not a technical question but an organizational one: which agencies issue binding targets, how targets are translated down administrative levels, which fiscal level funds construction, and who signs procurement contracts. The answer path follows central document 996 downward (NDRC Circular 2015-996) through to the "Procuring Entity" line in county-level procurement notices ^[1].

China's Institutional Architecture

In May 2015, nine central agencies (the National Development and Reform Commission, Office of the Central Commission for Comprehensive Social Management (zhongyang zongzhiban), Ministry of Science and Technology, Ministry of Industry and Information Technology, Ministry of Public Security, Ministry of Finance, Ministry of Human Resources and Social Security, Ministry of Housing and Urban-Rural Development, and Ministry of Transportation) jointly issued an Opinion on Strengthening Public Safety Video Surveillance Construction and Network Integration Work (NDRC Circular 2015-996). It set a unified goal: "full coverage, full sharing, full-time availability, complete control" ^[1].

Among the document's quantified targets for 2020, the four covering key public areas were: 100% coverage in key public areas, 100% of new and upgraded cameras in high definition, 100% network connectivity in key public areas, and 98% camera functionality ^[1]. These figures represent policy targets set at issuance to establish local performance baselines; they are not measurements of actual coverage. Public documents contain no nationwide implementation data to verify whether these targets were achieved or to what degree ^[10].

The financing clause was equally consequential: the document stipulates that construction, network integration, and maintenance costs for key public area video systems must be "included in this level's government fiscal budget." The central government sets targets but provides no central appropriation; provinces, cities, and counties fund construction and operations from their own fiscal budgets ^[1].

The Office of the Central Commission for Comprehensive Social Management on the signing agency list is not a government department but an institution within the Party's Political and Legal Affairs Commission system. Its inclusion signals that public video surveillance was embedded from inception in two administrative tracks: the government line handles project approval and construction, while the Party's comprehensive management line incorporates it as a performance accountability mechanism ^[1][8].

The 2018 Central Document No. 1, Opinion on Implementing the Rural Revitalization Strategy, further advanced this into rural areas: it required promotion of the Sharp Eyes Program in rural construction and mandated improved county-township-village comprehensive management centers with grid-based management systems ^[2].

In February 2025, the State Council issued the Public Safety Video Image Information System Management Regulations, the first administrative regulation to uniformly codify construction entities and management responsibilities for public video systems; the Ministry of Public Security issued accompanying implementation rules ^[3][4]. A decade of policy-driven and assessment-driven construction was thereby retrospectively given a regulatory framework.

How It Works

The two infrastructure lines divide responsibility: Skynet and Safe City programs use municipal public security bureaus as construction entities, with cameras linked to public security command centers; the Sharp Eyes Program is steered by the Party's comprehensive management system, extending into rural areas through county-township-village management centers, with the two lines organizationally separate under public security and Party management respectively ^[8][11][12].

From central document to county camera, each step has defined implementers: Circular 996 sets percentage targets and assigns funding to the local fiscal level; provincial governments issue implementation plans breaking down tasks; from 2016, the central comprehensive management office identified and promoted Sharp Eyes demonstration cities with replicable models; county governments and public security bureaus issue phased procurement notices and outsource operations through "service purchase" contracts ^[1][12][6].

This fiscal structure shapes local conduct: the central government appropriates nothing for cameras but sets percentage targets and incorporates them into comprehensive management evaluation; county treasuries absorb costs through phased construction and annual service purchases, transforming surveillance expansion into routine local budget items ^[1][8].

The Sharp Eyes name and early implementation model came from demonstration zones like Linyi, Shandong: surveillance feeds were piped into residents' set-top boxes to mobilize public participation in monitoring, drawing the name from the maxim that "the people's eyes are sharp." From 2016, the central comprehensive management office promoted this nationwide through demonstration city selection ^[12][10]. CSET researchers interpret this design as linking Maoist traditions of mass mutual surveillance with modern video recognition technology as a system of social control ^[9]; journalists and analysts tracking the program note that the "100% public space coverage" goal functions as a local performance lever ^[10].

Key Facts

Critical timeline nodes: Circular 996 establishes the program and sets 2020 targets in May 2015 ^[1]; Sharp Eyes demonstration city expansion nationwide begins in 2016 ^[12]; the 2018 Central Document No. 1 incorporates rural Sharp Eyes into rural revitalization ^[2]; the 2025 State Council Regulations establish unified construction and management responsibility ^[3].

On scale, ChinaFile's analysis of approximately 76,000 government procurement notices (2004 through May 2020) yields three figures: 2016 to 2020 procurement spending explicitly mentioning Sharp Eyes exceeded 14 billion yuan; in 2019 alone at least 998 counties procured surveillance equipment; surveillance procurement notice volume grew nearly 1,900 percent between 2010 and 2019 ^[8].

County notices themselves serve as organizational verification materials. In August 2022, Wenling City, Zhejiang, issued procurement documents for Skynet Phase II video image service: the procuring entity is Wenling Public Security Bureau, the three-year service budget is 63 million yuan, using equipment lease plus operations maintenance through service purchase ^[6]. This notice confirms three organizational facts: the Skynet line operates under the public security track; "Phase II" numbering corresponds to phased expansion; "service purchase" language corresponds to outsourced maintenance, converting one-time capital investment to annual fiscal spending.

In December 2024, Yuhang District, Hangzhou, released 2025 Sharp Eyes Program video surveillance service purchase procurement documents, with Yuhang District People's Government Zhongtai Subdistrict Office as procuring entity ^[5]. The appearance of a subdistrict office in the procuring entity field indicates that Sharp Eyes spending units have descended to township and subdistrict levels.

Xiangyin County, Hunan, in September 2021 announced a Sharp Eyes Program winning bid showing procurement for high-definition checkpoint capture, facial recognition, and vehicle capture systems, with the procuring entity Xiangyin County People's Government Office ^[7]. County-level Sharp Eyes procurement scope now encompasses facial and vehicle recognition capture equipment, not merely general video monitoring; the procuring entity being the government office rather than the public security bureau contrasts directly with the Wenling Skynet model.

Where these systems feed once built is best documented in Xinjiang, which provides the most complete technical evidence. Human Rights Watch's reverse engineering of a Xinjiang police application shows public video and checkpoint data connected to the Integrated Joint Operations Platform (IJOP), with camera location data participating in automatic generation of inspection and disposal orders ^[13]. The UN Office of the High Commissioner for Human Rights 2022 assessment determined that Xinjiang engages in large-scale arbitrary surveillance including through technical means ^[14]; the U.S. Treasury Department in December 2021 imposed investment restrictions on relevant Chinese companies citing the use of biometric surveillance technology for surveillance purposes ^[15].

Disputes and Evidence Limits

No reliable figure exists for total Chinese surveillance cameras. A 2019 IHS Markit report estimated roughly 770 million surveillance cameras worldwide at the time, with about 54 percent in China, and projected the global total would pass 1 billion by 2021, more than half of them in China ^[17]. Comparitech's current compilation instead cites a figure of approximately 700 million, which traces to a witness's written statement submitted to a March 2024 U.S. House hearing; the statement itself gives no data source ^[16]. These estimates employ incompatible statistical definitions: whether private cameras are included, whether counts cover registered or installed units, whether figures are projections or inventories. Standards vary across sources, and the figures come either from vendor-side market research or from unsourced hearing statements, with no official census behind any of them ^[16]. Only a range and definitional note can be stated; treating any single figure as fact exceeds what evidence supports.

"100% coverage" requires parallel treatment. Verifiable: Circular 996 inscribed 100 and 98 percent as 2020 policy targets, and local construction and procurement proceed using these targets as drivers ^[1][10]. Not verifiable: actual coverage rates. Public materials contain no nationwide implementation data; whether the targets were achieved or to what degree has no independent verification.

ChinaFile's 14 billion yuan figure is a lower-bound estimate: the count captures only projects explicitly marked "Sharp Eyes" in notices; surveillance procurement untagged as Sharp Eyes remains excluded ^[8].

Finally, evidence type boundaries exist. Procurement notices can verify organizational facts: procuring entity track affiliation, project phase number, expenditure structure; they cannot verify actual device online rates, actual data use, or use effectiveness. Judgments like "linking mass mobilization tradition with recognition technology" derive from research reports and investigative analysis of documents and should be distinguished from facts recorded in official documents themselves ^[9][10].

Our Position

Organizationally, a complete chain is verifiable: a nine-agency document sets percentage targets and assigns construction and maintenance costs to the local fiscal level ^[1]; the comprehensive management system advances the program to county-township-village through demonstration city selection and Central Document No. 1 ^[12][2]; county governments and public security bureaus implement through phased procurement and service purchase contracts ^[6][5][7]. Each link has public documentary support requiring no classified material access.

The mechanism's logic separates directive from cost: targets issue from central documents, money comes from local treasuries, evaluation occurs through the comprehensive management line. This explains the diffusion speed whereby nearly 1,000 counties entered procurement rolls in a single year ^[8]. Analysis stops at the organizational level: actual national coverage and device totals lack sufficient public evidence for resolution ^[16].

Sources

Policy texts: Opinion on Strengthening Public Safety Video Surveillance Construction and Network Integration Work (NDRC Circular 2015-996, nine agencies, 2015) ^[1]; Opinion on Implementing the Rural Revitalization Strategy (Central Document No. 1, 2018) ^[2]; Public Safety Video Image Information System Management Regulations (State Council, 2025) ^[3] and Ministry of Public Security implementation rules ^[4]. Verification method: use official government or ministry original texts, spotchecking agency signatures, target figures, and financing clauses.

Procurement notices: Wenling Skynet Phase II video image service procurement (Zhejiang, 2022) ^[6]; Yuhang District 2025 Sharp Eyes Program video surveillance service purchase notice (Hangzhou, 2024) ^[5]; Xiangyin County Sharp Eyes Program HD checkpoint and facial/vehicle capture system winning bid (Hunan, 2021) ^[7]. Verification method: observe procuring entity track affiliation, phase numbering, and "service purchase" language; county government pages are ephemeral; archive webpages simultaneously when citing.

Independent data and research: ChinaFile analysis based on approximately 76,000 procurement notices ^[8]; CSET report (Dahlia Peterson, 2020) ^[9]; OneZero reporting on Sharp Eyes (CSET republication, 2021) ^[10]; IPVM China Public Video Surveillance Guide ^[11]; China Digital Times series Part 1 (2019, including Linyi demonstration details) ^[12]; Human Rights Watch Xinjiang police app reverse engineering report (2019) ^[13]; UN Office of the High Commissioner for Human Rights Xinjiang assessment (2022) ^[14]; U.S. Treasury biometric surveillance sanctions announcement (2021) ^[15]; CNBC report on IHS Markit's 2019 surveillance camera forecast (2019) ^[17]; Comparitech camera estimate compilation ^[16].

Sources

1. Several Opinions on Strengthening Public Security Video Surveillance Construction, Networking and Application (NDRC Gaoji [2015] No. 996)primary-recordUnchecked
2. Opinions of the CPC Central Committee and the State Council on Implementing the Rural Revitalization Strategy (2018 Central Document No. 1)primary-recordUnchecked
3. Regulation on Public Security Video Image Information Systemsprimary-recordLive

   View captured page
4. MPS Rules for Public Security Video Information Systemsprimary-recordUnavailable
5. Yuhang District 2025 Sharp Eyes Video Surveillance Purchased-Services Project Tender Noticeprimary-recordUnchecked
6. Wenling City Skynet Phase II Video Image Services Procurement Tender Documents (Project No. WLCG-2022-13-02-GK)primary-recordUnchecked
7. Xiangyin County Sharp Eyes Project HD Checkpoint, Facial and Vehicle Capture System Government Procurement Award Noticeprimary-recordUnchecked
8. State of Surveillance: Government Documents Reveal New Evidence on China's Use of Surveillance Technologyinvestigative-reportingUnchecked
9. Designing Alternatives to China's Repressive Surveillance Stateacademic-researchUnchecked
10. China's "Sharp Eyes" Program Aims to Surveil 100% of Public Spaceinvestigative-reportingUnchecked
11. China Public Video Surveillance Guide: From Skynet to Sharp Eyestechnical-researchUnchecked
12. Sharper Eyes: Surveilling the Surveillers (Part 1)investigative-reportingUnchecked
13. China's Algorithms of Repression: Reverse Engineering a Xinjiang Police Apptechnical-researchLive

    View captured page
14. OHCHR Assessment of Human Rights Concerns in Xinjiangofficial-findingUnchecked
15. Treasury Sanctions on Biometric Surveillance Technologyofficial-findingLive

    View captured page
16. Surveillance Camera Statistics: Which Cities Have the Most CCTV Cameras?technical-researchUnchecked
17. CNBC on IHS Markit 2019 forecast: one billion surveillance cameras worldwide by 2021, more than half in China (770 million installed at the time, 54% in China)investigative-reportingUnchecked

×