In a place kept under wraps, John Honovich was on his laptop, methodically going through every link on a website for a conference on the other side of the world. Hikvision, the largest in the world security camera manufacturer, was hosting the event, the 2018 AI World Cloud Summit, in its hometown of Hangzhou, a city of about 10 million people not far from Shanghai. Honovich, the founder of a small trade publication that covered videos surveillance technology, he wanted to know what the latest equipment from Hikvision could do.
He zeroed in on a section of the conference agenda titled “Green, Peaceful, Relaxed” and found a description of an AI-powered system set up around Mount Tai, a historically sacred mountain in Shandong. Video showed Hikvision cameras trained on tourists climbing the thousands of stone steps leading up to the famous peak. Piano music was played as a narrator explained, in Mandarin with English subtitles, that the cameras were there “to identify all visitors and ensure everyone’s safety.” The video cut to a shot of a computer screen, and Honovich paused. He saw a magnified view of a visitor’s face. Beneath it was data that the camera’s AI had inferred. Honovich downloaded the video and took screenshots of the computer screen for safekeeping.
Later, with the help of a translator, he examined each piece of text on that screen. One set of characters, the translator explained, suggested that each visitor would automatically fall into categories: age, gender, wearing glasses, smiling. When Honovich pointed to the fifth category and asked, “What is this?” the translator replied: “minority”. Honovich pressed: “Are you sure?” The translator confirmed that there was no other way to read it.
Honovich was surprised. In his many years in the industry, he had never seen a surveillance company willing to automatically detect racial minorities. He found the feature completely unethical, and immediately wondered how China could use it against the Uyghur people, a largely Muslim minority ethnic group, in Xinjiang province. Honovich had seen reports in the west of Uyghurs subjected to restrictive surveillance and mass arrests. By clicking through to the AI Summit website, Honovich was unable to learn whether the Chinese authorities were using this technology to oppress minorities, but saw that the danger merged. She quickly wrote an article about Hikvision’s ethnicity detection technology, including video, screenshots, and a non-commentary comment from the company, and published on the website of IPVM, the trade publication he had founded.
He discussed the discovery with one of IPVM’s reporters, Charles Rollet, a Frenchman who lives outside the US and also keeps its location secret. Rollet had written about how Hikvision and Dahua, China’s second-biggest video surveillance manufacturer, were making huge profits from the government’s work in Xinjiang. Rollet had newspaper experience, and though he was 25, he spoke like an inky newscaster twice his age, all “scoops” and “calling out abuse” and “hard-hitting news.” By trawling publicly available materials online, Rollet learned that Hikvision had reached an agreement to build a facial recognition system to cover a Xinjiang county, including a “re-education” center and some of its mosques, and a contract to install videoconferencing systems in mosques, presumably so attendees could view government-broadcast sermons. Dahua won the biggest contract: $686 million to build camera-equipped police stations in another part of Xinjiang. The agreements specified that companies would install these systems, run them for several years, and then turn them over to the government. In many aspects of the government’s video surveillance in Xinjiang, Rollet reportedthe two companies were “deeply involved”.
Hikvision and Dahua cameras have also been hung in homes, businesses and public buildings in the US and much of the world. Security system installers enthusiastically sold large numbers of cheap cameras. Global financial institutions such as Fidelity International and Norway’s sovereign wealth fund were eager investors in fast-growing and profitable Chinese companies. US chip giants Intel and Nvidia sold them silicon to power their facial recognition.
All that would change soon. Over the next several years, IPVM writers unearthed one damning detail after another in Chinese surveillance team. His scoops would end up influencing national politics, changing the fortunes of those companies and putting the reporters themselves squarely on the front lines of the cold war between the United States and China.
I met for the first time Honovich on a summer day in New York, in Brooklyn’s Marine Park, not far from where he grew up. There are no photos of him on the IPVM site or on his LinkedIn, a decision I would later understand. He is a man of small build, with glasses over his closely set eyes and a boyish face. We sat down at a small table with an inlaid chessboard, and Honovich, dressed in shorts and a T-shirt, began talking to me about the surveillance industry in rapid-fire sentences.
—————————————————-
Source link