Recently uncovered evidence reveals that Chinese telecoms giant Huawei and other Chinese tech firms have filed patents for artificial intelligence technology that detects Uyghurs and other ethnic minorities in China. Now the tech firms are attempting to backpedal and downplay the gravity of the revelations, some claiming it’s all a “misunderstanding”.
[[post-object type=”divider” /]]
IPVM, a video surveillance research firm based in the United States has exposed evidence showing that Huawei, Chinese facial recognition startups Megvii and SenseTime, Alibaba, Baidu, and several other major Chinese tech companies filed patent applications for facial recognition and other AI-based technology that has the ability to classify individuals based on several different attributes, including ethnicity. Though many of the patents only mentioned “ethnicity” or “race” as being one of the classifiable attributes, the Huawei, Megvii, and SenseTime patents each explicitly mentioned the capacity to detect Uyghurs specifically.
Huawei’s patent application for AI technology for the “identification of pedestrian attributes” specifically mentions that “the attributes of the target object can be gender (male, female), age (such as teenagers, middle-aged, old), race (Han, Uyghur)… etc.” Yet in response to the revelation of the patent, Huawei claims that “identifying individuals’ race was never part of the research and development project,” adding that “it should never have become part of the application and we are taking proactive steps to amend it.”
A patent filed in June of 2019 by Chinese facial recognition startup Megvii for a “portrait retrieval method and device” that “can also directly connect to the facial recognition that has been built by the public security organ” mentions “ethnicity classification standards” that can be “divided according to Han, non-Han and unknown, or according to Han, Uyghur, non-Han, non-Uyghur, and unknown.” But, according to Megvii, the language used in the patent is “open to misunderstanding.”
Apparently, Megvii expects the public to believe that in reality, “the patent application pertains to technology to re-label images based on existing attributes provided by third parties, where some of them might have been labeled incorrectly,” and that the “functionality, which re-labels images using such attributes as age, gender and ethnicity, is in no way an intention to develop ethnic identification solutions” even though the patent explicitly states that the technology can differentiate between Uyghurs and non-Uyghurs and can be integrated with facial recognition systems used for public security.
Despite denying it had any ill intentions with respect to using AI tech to detect Uyghurs, Megvii has reportedly vowed to withdraw the patent following IPVM’s revelations.
Similarly, SenseTime, which happens to be China’s largest facial recognition startup, applied for a patent for a “method and device for retrieving images”. According to the SenseTime patent, the technology in question has the capacity to “divide…ethnicities according to actual needs“, and “can be divided according to Han, non-Han and unknown, or according to Han, Uyghur, non-Han, non-Uyghur, and unknown”. SenseTime was contrite in its response to IPVM’s revelations but maintained that its technology was intended to include “facial recognition of all ethnicities without prejudice,” adding that “the reference to Uyghurs is regrettable” and that “it was neither designed nor intended in any way to discriminate, which is against our values”. The company claims that the reference to Uyghurs was only to serve as to illustrate an example of the types of attributes the AI algorithm is capable of recognizing.
Of course, each company exposed by IPVM’s research responded in an equally predictable manner in deflecting and downplaying its participation in furthering the plight of an already terribly oppressed minority group. The uncovered patent applications speak for themselves, and no amount of PR acrobatics will convince the public that the companies’ intentions were entirely benign.
It is widely asserted that the Chinese government is active in using intrusive surveillance methods to monitor the Uyghur population in China to help send over a million ethnic Uyghurs to forced-labor ‘re-education’ camps. And Chinese tech firms like Huawei, Megvii, SenseTime, and others are actively developing AI and facial recognition solutions to help Beijing carry out its contemptible human rights abuses against the Uyghur ethnic minority.
Facial recognition technology already has the potential to be abused heavily and to infringe on one’s privacy, but when the technology is leveraged for purposes related to ethnic detection and can aid authoritarian governments in persecuting individuals based on their race, the dangers and consequences are elevated to an entirely new level. Ultimately, the most frightening element of IPVM’s revelations is how utterly mainstream and common it is for massively influential Chinese tech firms to engineer technological solutions that encourage exactly that.