Another Chinese facial recognition provider is coming under fire for developing technology that would allow the police to identify members of the country’s minority Uighur Muslim population. The news comes only a few short months after IPVM investigators uncovered a similar utility in technology developed by Megvii and Huawei.
This time around, Dahua is the company facing international scrutiny. As before, IPVM was responsible for unearthing the new ethnic identification evidence. The organization investigates computer surveillance systems, and provided the Los Angeles Times with screenshots of Dahua product support documents that highlight the ability to identify people of various ethnicities. According to those documents, Dahua’s technology will send real-time notifications to police whenever it spots a member of a targeted ethnic group, while private companies can use the system to categorize the faces of potential customers based on their race.
It’s unclear if the ethnic recognition tool has ever been deployed in a real-world setting. For its part, Dahua responded to the allegations with a statement that claimed that it “does not sell products that feature [an] ethnicity-focused recognition function.” However, the company quietly deleted code that pointed to the ethnic identification tool when an engineer first called attention to it in November. That code was part of a software development kit that the company had posted to Github’s public code-sharing platform.
Dahua, Megvii, and Huawei are all on a US entity list that prevents them from purchasing American products, though that does not prevent them from selling their own products to interested US customers. Meanwhile, privacy advocates warned that Dahua’s technology could help to perpetuate racial bias if deployed in the US. Dahua currently maintains US sales offices in Irvine and Houston, and has more than 80 contracts in California alone.
The Council of Europe recently asked the European Union to ban facial recognition systems designed to identify members of specific demographic groups. Multiple groups have stressed that the elimination of bias is a key priority when developing ethical facial recognition systems.
Source: Los Angeles Times
February 9, 2021 – by Eric Weiss