“You have been spotted as having participated in acts of violence in the Al-Aqsa Mosque,” it read in Arabic. “We will hold you accountable.”
Ramlawi, then 19, was among hundreds of people who civil rights attorneys estimate got the text last year, at the height of one of the most turbulent recent periods in the Holy Land. Many, including Ramlawi, say they only lived or worked in the neighborhood, and had nothing to do with the unrest. What he didn’t know was that the feared internal security agency, the Shin Bet, was using mass surveillance technology mobilized for coronavirus contact tracing, against Israeli residents and citizens for purposes entirely unrelated to COVID-19.
In the pandemic’s bewildering early days, millions worldwide believed government officials who said they needed confidential data for new tech tools that could help stop coronavirus’ spread. In return, governments got a firehose of individuals’ private health details, photographs that captured their facial measurements and their home addresses.
Now, from Beijing to Jerusalem to Hyderabad, India, and Perth, Australia, The Associated Press has found that authorities used these technologies and data to halt travel for activists and ordinary people, harass marginalized communities and link people’s health information to other surveillance and law enforcement tools. In some cases, data was shared with spy agencies. The issue has taken on fresh urgency almost three years into the pandemic as China’s ultra-strict zero-COVID policies recently ignited the sharpest public rebuke of the country’s authoritarian leadership since the pro-democracy protests in Tiananmen Square in 1989.
For more than a year, AP journalists interviewed sources and pored over thousands of documents to trace how technologies marketed to “flatten the curve” were put to other uses. Just as the balance between privacy and national security shifted after the Sept. 11 terrorist attacks, COVID-19 has given officials justification to embed tracking tools in society that have lasted long after lockdowns.
“Any intervention that increases state power to monitor individuals has a long tail and is a ratcheting system,” said John Scott-Railton, a senior researcher at the Toronto-based internet watchdog Citizen Lab. “Once you get it, is very unlikely it will ever go away.”
In China, the last major country in the world to enforce strict COVID-19 lockdowns, citizens have been required to install cell-phone apps to move about freely in most cities. Drawing from telecommunications data and PCR test results, the apps produce individual QR codes that change from green to yellow or red, depending on a person’s health status.
The apps and lockdowns are part of China’s sweeping pandemic prevention policies that have pushed the public to a breaking point. When an apartment fire in Urumqi last month left at least 10 dead, many blamed zero-tolerance COVID policies. That sparked demonstrations in major cities nationwide, the largest display of defiance in decades, after which the government announced it would only check health codes in “special places,” such as schools, hospitals and nursing homes.
Last week, the government went further, saying it would shut down a national-level health code to ease travel between provinces. But cities and provinces have their own codes, which have been more dominant. In Beijing last week, restaurants, offices, hotels and gyms were still requiring local codes for entry.
Over the past few years, Chinese citizens have needed a green code to board domestic flights or trains, and in some cities even to enter the supermarket or to get on a bus. If they were found to have been in close contact with someone who tested positive for COVID-19, or if the government imposed a local quarantine, the code would turn red, and they were stuck at home.
There’s evidence that the health codes have been used to stifle dissent.
This story, supported by the Pulitzer Center on Crisis Reporting, is part of an ongoing Associated Press series, “Tracked,” that investigates the power and consequences of decisions driven by algorithms on people’s everyday lives.
In early September, former wealth manager Yang Jiahao bought a train ticket to Beijing, where he planned to lodge various complaints with the central government. The night before, a woman he described as a handler invited him to dinner. Handlers are usually hired by state security as part of “stability maintenance” operations and can require people to meet or travel when authorities worry they could cause trouble. Yang had a meal with the handler, and the next morning Guangzhou health authorities reported a COVID-19 case less than a kilometer from where they dined, he said.
Based on city regulations, Yang’s code should have turned yellow, requiring him to take a few COVID tests to show he was negative.
Instead, the app turned red, even though tests showed that he didn’t have COVID. Yang was ordered to quarantine and a paper seal was placed on his door.
“They can do whatever they want,” he said.
An officer at the Huangcun station of the Guangzhou police referred comment to city-level authorities on Yang’s case, saying he required proof that the caller was from the AP. Guangzhou’s Public Security Bureau and the city’s Center for Disease Control and Prevention did not respond to faxed requests for comment.
In another show of how the apps can control lives, in June, a group of bank customers were effectively corralled by the health codes when they tried going to Henan’s provincial capital in Zhengzhou to protest being unable to access their online bank accounts.
A notice said the problem was due to a system upgrade. But the customers soon found out the real reason: a police investigation into stockholders in the parent bank had rendered 40 billion yuan in funds inaccessible, according to local media reports. Frustrated after months of complaints, a group of customers decided to hold a protest in Zhengzhou at the provincial banking commission.
Customer Xu Zhihao uploaded his itinerary to get the Henan province health code after he tested negative for COVID-19 in his coastal city of Tianjin, just south of Beijing. As he got off the train in Zhengzhou, Xu was asked to scan his QR code at the station, and immediately it turned red. The train station employee called security and took him to a police booth.
Xu said police took him to the basement to quarantine. Three other people joined him, and all four realized that they had come to get their money back.
“They had set the net in place, waiting for us,” Xu said.
From a group chat, Xu and others learned that many protesters had met a similar fate, at the high-speed rail train station, at the airport and even on the highway. A government inquiry later found that red codes were given to 1,317 people, many of whom had planned to protest.
China’s National Health Commission, which has led the COVID response, did not reply to a fax requesting comment. The Henan provincial government did not respond either.
Even after China ends lockdowns, some dissidents and human rights activists predict the local-level health codes will stay on as a technological means of social control. Early on, provinces didn’t share data, but in the past few years, that has changed.
Some provincial governments have created local apps that can link health, location and even credit information, which leaves open the possibility for these apps or the national databases they draw from to be used to monitor people in the future, according to an AP review of procurement documents, research and interviews. Xu and Yang, for instance, were both stopped in their tracks by local health codes.
In February, police in northeastern Heilongjiang province sought to upgrade their local health code so they could search PCR test results for anyone in China, in real time, according to procurement documents provided exclusively by ChinaFile, a digital magazine published by the Asia Society. A company whose parent is government-owned won the non-competitive bid to connect that app to a national database of PCR data run by the State Council, China’s Cabinet, fulfilling a national directive, the documents show. The same company, Beijing Beiming Digital Technology, also claims on its website that it has developed more than 30 pandemic apps.
“It’s the governance model, the philosophy behind it is to strengthen social control through technology. It’s strengthened by the health app, and it’s definitely going to stay after COVID is over,” said Yaqiu Wang, a senior researcher with Human Rights Watch. “I think it’s very, very powerful.”
“THERE ARE TWO SETS OF LAWS”
In Jerusalem’s Old City, tourists sipping fresh pomegranate juice, worshippers and locals taking a shortcut home are all monitored by Israeli security forces holding automatic weapons. The labyrinth of cavernous pathways is also lined with CCTV cameras and what authorities have described as “advanced technologies.”
After clashes in May 2021 at the Al-Aqsa Mosque helped trigger an 11-day war with Hamas militants in the Gaza Strip, Israel experienced some of the worst violence in years. Police lobbed stun grenades into the disputed compound known to Jews as the Temple Mount, home to Al-Aqsa, Islam’s third-holiest site, as Palestinian crowds holed up inside hurling stones and firebombs at them.
By that time, Israelis had become accustomed to police showing up outside their homes to say they weren’t observing quarantine and knew that Israel’s Shin Bet security agency was repurposing phone surveillance technology it had previously used to monitor militants inside Palestinian territories. The practice made headlines at the start of the pandemic when the Israeli government said it would be deployed for COVID-19 contact tracing.
A year later, the Shin Bet quietly began using the same technology to send threatening messages to Israel’s Arab citizens and residents whom the agency suspected of participating in violent clashes with police. Some of the recipients, however, simply lived or worked in the area, or were mere passers-by.
Ramlawi’s coffeeshop sits in the ornate Cotton Merchant’s Market outside the mosque compound, an area lined with police and security cameras that likely would have identified the barista had he participated in violence.
Although Ramlawi deleted the message and hasn’t received a similar one since, he said the thought of his phone being used as a monitoring tool still haunts him.
“It’s like the government is in your bag,” said Ramlawi, who worries that surveillance enabled to stop COVID-19 poses a lasting menace for east Jerusalem residents. “When you move, the government is with you with this phone.”
The Shin Bet’s domestic use of the technology has generated an uproar over privacy and civil liberties within Israel, as well as questions about its accuracy. The Ministry of Communications, which oversees Israel’s telecommunications companies, refused a request seeking further details submitted for the AP by the Movement for Freedom of Information, a nonprofit that frequently works with media organizations.
Gil Gan-Mor, an attorney with the nonprofit Association for Civil Rights in Israel, estimates that hundreds of Arabs in Jerusalem received the threatening message during the unrest and said the mass text message blast was unprecedented.
“You cannot just say to people, ‘We are watching you … and we will get revenge,” he said. “You cannot use this tool to frighten people. If you have something against someone, you can put them on trial.’”
After Gan-Mor’s organization sued, Shin Bet made no apologies.
“There was a clear security need to send an urgent message to a very large number of people, all of whom had a credible suspicion of being involved in performing violent crimes,” the agency said in a legal filing last year. The filing, signed by “Daniella B.,” the Shin Bet’s legal adviser for the Jerusalem district, also acknowledged that “lessons were learned.”
In February, Israel’s attorney general upheld the continued use of the technology, saying it was a legitimate security tool, while acknowledging glitches in the system and that messages were distributed to a small number of unintended targets. Israel’s Supreme Court is now reviewing the matter.
Sami Abu Shehadeh, a former Arab lawmaker who served in Israel’s parliament at the time Shin Bet sent its warning texts, said the messages demonstrate the broader struggles of Israel’s 20% Arab minority.
“The state does not deal with us as citizens,” he said. “There are two sets of laws — one for Jews and one for Arabs.”
‘360 DEGREE SURVEILLANCE’
Technologies designed to combat COVID-19 were redirected by law enforcement and intelligence services in other democracies as governments expanded their digital arsenals amid the pandemic.
In India, facial recognition and artificial intelligence technology exploded after Prime Minister Narendra Modi’s right-wing Hindu nationalist Bharatiya Janata Party swept into power in 2014, becoming a tool for police to monitor mass gatherings. The country is seeking to build what will be among the world’s largest facial recognition networks.
As the pandemic took hold in early 2020, state and central governments tasked local police with enforcing mask mandates. Fines of up to $25, as much as 12 days’ pay for some laborers and unaffordable for the nearly 230 million people estimated to be living in poverty in India, were introduced in some places.
In the south-central city of Hyderabad, police started taking pictures of people flaunting the mask mandate or simply wearing masks haphazardly.
Police Commissioner C.V. Anand said the city has spent hundreds of millions of dollars in recent years on patrol vehicles, CCTV cameras, facial recognition and geo-tracking applications and several hundred facial recognition cameras, among other technologies powered by algorithms or machine learning. Inside Hyderabad’s Command and Control Center, officers showed an AP reporter how they run CCTV camera footage through facial recognition software that scans images against a database of offenders.
“When (companies) decide to invest in a city, they first look at the law-and-order situation,” Anand said, defending the use of such tools as absolutely necessary. “People here are aware of what the technologies can do, and there is wholesome support for it.”
By May 2020, the police chief of Telangana state tweeted about his department rolling out AI-based software using CCTV to zero-in on people not wearing masks. The tweet included photos of the software overlaying colored rectangles on the maskless faces of unsuspecting locals.
More than a year later, police tweeted images of themselves using hand-held tablets to scan people’s faces using facial recognition software, according to a post from the official Twitter handle of the station house officer in the Amberpet neighborhood.
Police said the tablets, which can take ordinary photographs or link them to a facial recognition database of criminals, were a useful way for officers to catch and fine mask offenders.
“When they see someone not wearing a mask, they go up to them, take a photo on their tablet, take down their details like phone number and name,” said B Guru Naidu, an inspector in Hyderabad’s South Zone.
Officers decide who they deem suspicious, stoking fears among privacy advocates, some Muslims and members of Hyderabad’s lower-caste communities.
“If the patrolling officers suspect any person, they take their fingerprints or scan their face – the app on the tablet will then check these for any past criminal antecedents,” Naidu said.
S Q Masood, a social activist who has led government transparency campaigns in Hyderabad, sees more at stake. Masood and his father-in-law were seemingly stopped at random by police in Shahran market, a predominantly Muslim area, during a COVID-19 surge last year. Masood said officers told him to remove his mask so they could photograph him with a tablet.
“I told them I won’t remove my mask. They then asked me why not, and I told them I will not remove my mask.” He said they photographed him with it in place. Back home, Masood went from bewildered to anxious: Where and how was this photo to be used? Would it be added to the police’s facial recognition database?
Now he’s suing in the Telangana High Court to find out why his photo was taken and to limit the widespread use of facial recognition. His case could set the tone for India’s growing ambition to combine emerging technology with law enforcement in the world’s largest democracy, experts said.
India lacks a data protection law and even existing proposals won’t regulate surveillance technologies if they become law, said Apar Gupta, executive director of the New Delhi-based Internet Freedom Foundation, which is helping to represent Masood.
Police responded to Masood’s lawsuit and denied using facial recognition in his case, saying that his photograph was not scanned against any database and that facial recognition is only used during the investigation of a crime or suspected crime, when it can be run against CCTV footage.
In two separate AP interviews, local police demonstrated both how the TSCOP app carried by police on the street can compare a person’s photograph to a facial recognition database of criminals, and how from the Command and Control Center police can use facial recognition analysis to compare stored mugshots of criminals to video gathered from CCTV cameras.
Masood’s lawyers are working on a response and awaiting a hearing date.
Privacy advocates in India believe that such stepped-up actions under the pandemic could enable what they call 360 degree surveillance, under which things like housing, welfare, health and other kinds of data are all linked together to create a profile.
The contract shows data broker Cuebiq provided a “device ID,” which typically ties information to individual cell phones. The CDC also could use the information to examine the effect of closing borders, an emergency measure ordered by the Trump administration and continued by President Joe Biden, despite top scientists’ objections that there was no evidence the action would slow the coronavirus.
CDC spokeswoman Kristen Nordlund said the agency acquired aggregated, anonymous data with extensive privacy protections for public health research, but did not address questions about whether the agency was still using the data. The CDC could still access aggregate, county-level mobile phone data through October, several months after its contract ended, Cuebiq spokesman Bill Daddi said on Tuesday. He added that before then, the CDC also could have analyzed privacy-protected individual mobile phone data, but chose not to.
For Scott-Railton, that sets a dangerous precedent.
“What COVID did was accelerate state use of these tools and that data and normalize it, so it fit a narrative about there being a public benefit,” he said. “Now the question is, are we going to be capable of having a reckoning around the use of this data, or is this the new normal?”
Former AP video journalist Rishabh R. Jain contributed to this report from Hyderabad, India. AP staffers Lori Hinnant contributed from Paris; María Verza from Mexico City; Astrid Suárez from Bogotá, Colombia; Edna Tarigan from Jakarta, Indonesia; Tong-hyung Kim from Seoul, South Korea; and Eileen Ng from Singapore. Daria Litvinova and retired Associated Press Afghanistan and Pakistan Bureau Chief Kathy Gannon also contributed. Deputy Editor of The Mail & Guardian Athandiwe Saba assisted from Johannesburg. Burke reported from San Francisco; Federman from Jerusalem; McGuirk from Canberra, Australia; Pathi from Hyderabad, India; and Wu from Taipei, Taiwan.
This reporting was produced in collaboration with researcher Avani Yadav with support from the Human Rights Center Investigations Lab at the University of California, Berkeley. It was partially supported by the Starling Lab for Digital Integrity, co-founded by the University of Southern California and Stanford University, where Burke was a journalism fellow.