Huawei Reportedly Tested a ‘Uighur Alarm’ to Track Chinese Ethnic Minorities With Facial Recognition

The system also identifies information such as age and sex

Dave Gershgorn
OneZero
Published in
2 min readDec 8, 2020

--

Photo: Greg Baker/AFP via Getty Images

Chinese tech giants Huawei and Megvii have allegedly tested software that could identify Uighurs, an ethnic minority in China, according to a new report from the Washington Post and video surveillance trade publication IPVM.

The system being tested tried to identify whether a person was Uighur but also information such as their age and sex. If the system detected a Uighur person, it could notify government authorities with a “Uighur alarm.” The system relies on Huawei’s cameras, cloud computing servers, and other hardware plus Megvii’s facial-recognition algorithms.

Uighurs have faced increasing surveillance and incarceration in China, from the mass collection of DNA to allegedly being forced into more than 250 detention centers.

Huawei and Megvii’s technology is not new. Last year IPVM identified more than a dozen state-run projects that use A.I. to try and detect Uighurs. Some of this research has even been done publicly, like a 2018 paper mentioned by the Post that specifically tried to differentiate Uighur face characteristics from people of Korean and Tibetan descent.

However, the companies that typically supply this technology specialize in security cameras and surveillance tech, like Dahua and Hikvision, instead of mainstream tech companies. Dahua, Hikvision, and Megvii have all been sanctioned by the U.S. government for supplying this technology.

Read the story from the Washington Post for more information. I also suggest following BuzzFeed News reporter Megha Rajagopalan. She’s covered the topic extensively from within China, and you can even read about her work in an interview on OneZero.

--

--

Dave Gershgorn
OneZero

Senior Writer at OneZero covering surveillance, facial recognition, DIY tech, and artificial intelligence. Previously: Qz, PopSci, and NYTimes.