By Eduardo Baptista

BEIJING, (Reuters) – Hundreds of Chinese companies have developed software that sorts data on residents using artificial intelligence. This was in response to high demands from authorities looking to improve their surveillance tools. A Reuters review of documents reveals this.

Reuters has found that dozens of Chinese entities have purchased this software over the last four years, according to 50 documents. This technology is an improvement on the existing software that simply collects data and leaves it up to individuals to organize.

“The system has the ability to learn independently and can optimize the accuracy of file creation as the amount of data increases. (Faces that are) partially blocked, masked, or wearing glasses, and low-resolution portraits can also be archived relatively accurately,” according to a tender published in July by the public security department of Henan, China’s third-largest province by population.

Henan’s Department of Public Security did not respond to inquiries about the system and it’s uses.

New software enhances Beijing’s surveillance approach. China’s current surveillance system can gather data about individuals but law enforcement agencies and other users are left with the task of organizing it.

Another limitation of current surveillance software is its inability to connect an individual’s personal details to a real-time location except at security checkpoints such as those in airports, according to Jeffrey Ding, a postdoctoral fellow at Stanford’s Center for International Security and Cooperation.

Mareike Ohlberg (a senior fellow at The German Marshall Fund based in Berlin) said that one person and one file is a method of sorting information which makes it easier for individuals to be tracked.

China’s Department of Public Security oversees the regional police authorities. It did not reply to my request for information about one individual, one file, and its surveillance. Ten bids were also opened by Chinese Communist Party organizations responsible for legal and political affairs. China’s Central Political and Legal Affairs Commission refused to comment.

Three industry professionals interviewed by Reuters said that the tenders being examined represent a small fraction of efforts made by Chinese police and Party units to improve surveillance networks through tapping into big data and artificial intelligence.

According to documents from the government, certain users of the software, including schools, wanted to see who was outside their boundaries.

Most, including police units from Ngawa Prefecture in Sichuan’s southwestern Sichuan province, which is dominated by Tibetans ordered the software for security reasons. Ngawa’s tender described the software as “maintaining social stability and political security among the people.”

The Ngawa department of Public Security did not respond when asked.

Beijing claims its surveillance is vital to fighting crime. It has also been crucial in its fight against COVID-19. Human Rights Watch, a group of human rights activists, claims that China is creating a surveillance system that violates privacy and targets unfairly certain groups such as the Uyghur Muslims minority.

The Reuters Review shows that at least 50 tenders have been opened by local authorities in China, in high-populated areas like Beijing, and in underdeveloped regions such as Gansu in the past four years. 32 of these were open for bidding in 2020. A Reuters review revealed that 22 companies offer this software including Sensetime (Huawei, Megvii), Cloudwalk and Dahua as well as the cloud division at Baidu (NASDAQ:).

Sensetime has declined to comment. Requests for comments from Megvii and Cloudwalk as well as Dahua and Baidu’s cloud division were not answered.

Huawei stated in a statement, that one partner developed the smart-city platform’s one person/one file application. Huawei declined to comment about the patent applications.

According to the company, “Huawei is not known for developing or selling applications targeted at any one group.”

Reuters’ reviews covered 22 of China’s main administrative divisions as well as all levels of provincial government from the regional public security offices to Party office for a particular neighbourhood.

These new systems will make sense out of the huge amount of data these entities hold. Complex algorithms and machine learning are used to create customized files that can be used by individuals. Files are automatically updated as software sorts data.

Implementation can be complicated by a wide variety of issues. According to three AI experts and surveillance specialists, bureaucracy or even cost can cause a fractured and disjointed national network.

Reuters discovered announcements of successful bids in more than 50% of 50 procurement documents it analysed. These auctions had values between a few thousand and nearly 200 million yuan.

SYSTEM UPGRADE

China covered its cities in surveillance cameras during a campaign that it called “sharp eyes”. It is now trying to replicate this strategy across the country’s rural regions. Around this time the “one file, one person” software was developed and adopted.

Ohlberg, the researcher, said the earliest mention she had seen of one person, one file was from 2016, in a 200-page surveillance feasibility study by Shawan county in Xinjiang, for acquiring a computer system that could “automatically identify and investigate key persons involved in terrorism and (threatening social) stability.” A Shawan county official declined to comment.

Meng Jianzhu (China’s domestic security head at that time), wrote in a journal state-run that big data was key for identifying crime patterns and trends. Li Ziqing was the then-director of Research Center for Biometrics and Security Technology of Chinese Academy of Sciences. He spoke to industry leaders two years later about the system. Li served as chief scientist for AuthenMetric in Beijing, a facial recognition company. Neither AuthenMetric or the research centre responded to inquiries for comment.

“The ultimate core technology of big data’s (application to) security is one person, one file,” Li said in the 2018 speech at an AI forum in Shenzhen, according to a transcript of the speech published by local media and shared on AuthenMetric’s WeChat public account.

Meng was the leader of 2016’s Party’s Political and Legal Affairs Commission. Meng declined to comment. Meng couldn’t be reached for comment. Li didn’t respond to our request for comment.

This industry saw rapid growth. By 2021 Huawei and Sensetime had submitted patent applications to the World Intellectual Property Organization. These were for image clustering and file archiving.

A 2021 Huawei patent application for a “person database partitioning method and device” that mentioned one person, one file said that “as smart cameras become more popular in the future, the number of captured facial images in a city will grow to trillions per year”.

    SAFE CITIES

The 50 tenders Reuters examined provide varying details on the use of the software.

“One person, one folder” was used by some to refer to a single entry from a list containing items required for surveillance systems. Some others gave more detailed descriptions.

9 of the tenders stated that the software could be integrated with facial recognition technology. This can, according to the documents, determine if a person is Uyghur and connect to police early warning systems.

One tender published in February 2020 by a Party organ responsible for an area in the southeastern island province of Hainan, for instance, sought a database of Uyghur and Tibetan residents to facilitate “finding the information of persons involved in terrorism.”

A request to comment was not received by the Hainan authorities

More than a dozen tenders mention the need to combat terrorism and “maintain stability”, a catch-all term that human rights activists say is often used to mean repressing dissent.

Four tenders stated that the software must be capable of pulling information from an individual’s social networks accounts. A majority of tenders stated that the software could be used for analysis and compilation of personal data such as marital status, family, friends, vehicles, shopping habits, and vehicle records.