‘One Person, One File’ systems using AI to automatically identify individuals and classify data spreading rapidly in China, claims report
Artificial intelligence-driven systems capable of automatically organising vast amounts of surveillance data collected on individuals are proliferating rapidly in China, according to a report.
The systems, known as “One Person, One File”, have been built by dozens of Chinese firms and acquired by dozens of entities in the country, including law enforcement and government bodies, Reuters reported.
The technology, which can be linked to facial-recognition images captured by surveillance cameras, adds to the concerns of human rights activists over the potential for misuse of pervasive state surveillance systems.
While existing systems typically require humans to organise data collected on individuals, One Person, One File systems are designed to automatically identify people and collate information on them.
Reuters said its report was based on an analysis of 50 publicly available procurement documents and other data.
Human rights activists say such systems can be used to stifle dissent, particularly from ethnic minorities.
Nine of the tenders analysed by Reuters indicated the software being acquired would be used with facial recognition to identify whether a passerby belonged to the minority Uyghur ethnic group, connecting to early-warning systems for police and creating archives of Uyghur faces.
A tender published in February 2020 by a Communist Party organ in the southeastern island province of Hainan sought a database of Uyghur and Tibetan residents to facilitate “finding the information of persons involved in terrorism”, according to Reuters.
Ngawa prefecture in Sichuan province, which is mainly populated by Tibetans, said its system would be used for “maintaining political security, social stability and peace among the people”.
A tender published last July by the public security department of Henan province indicated its system would be able to identify and archive faces that are “partially blocked, masked, or wearing glasses, and low-resolution portraits”.
The first patent for One Person, One File was filed four years ago, since which time local authorities across China have filed at least 50 tenders for such systems, 32 of which were opened in 2021, according to Reuters.
It found announcements of successful bids for more than half of the 50 procurement documents analysed.
At least four of the tenders said the software should be able to pull information from the individual’s social media accounts, with half saying it would be used to compile and analyse personal details such as relatives, social circles, vehicle records, marriage status and shopping habits, Reuters said.
The news agency said 22 tech companies, including Sensetime, Huawei, Megvii, Cloudwalk, Dahua, and the cloud division of Baidu now offer such software.
Huawei said it “does not develop or sell applications that target any specific group of people”.
Chinese officials declined to comment for Reuters’ report.
Last year China’s embassy in the UK told the BBC programme Panorama that there was no basis to reports of facial recognition technology targeting ethnic minorities.
“There is no so-called facial recognition technology featuring Uyghur analytics whatsoever,” the embassy said at the time.