AI emotion-detection software tested on Uyghurs

The police state becomes even more sophisticated:

A camera system that uses AI and facial recognition intended to reveal states of emotion has been tested on Uyghurs in Xinjiang, the BBC has been told.

A software engineer claimed to have installed such systems in police stations in the province.

A human rights advocate who was shown the evidence described it as shocking.

The Chinese embassy in London has not responded directly to the claims but says political and social rights in all ethnic groups are guaranteed.

Xinjiang is home to 12 million ethnic minority Uyghurs, most of whom are Muslim.

Citizens in the province are under daily surveillance. The area is also home to highly controversial “re-education centres”, called high security detention camps by human rights groups, where it is estimated that more than a million people have been held. 

Beijing has always argued that surveillance is necessary in the region because it says separatists who want to set up their own state have killed hundreds of people in terror attacks.

The software engineer agreed to talk to the BBC’s Panorama programme under condition of anonymity, because he fears for his safety. The company he worked for is also not being revealed. 

But he showed Panorama five photographs of Uyghur detainees who he claimed had had the emotion recognition system tested on them.

image captionData from the system purports to indicate a person’s state of mind, with red suggesting a negative or anxious state of mind

“The Chinese government use Uyghurs as test subjects for various experiments just like rats are used in laboratories,” he said.

And he outlined his role in installing the cameras in police stations in the province: “We placed the emotion detection camera 3m from the subject. It is similar to a lie detector but far more advanced technology.”

He said officers used “restraint chairs” which are widely installed in police stations across China.

“Your wrists are locked in place by metal restraints, and [the] same applies to your ankles.”

He provided evidence of how the AI system is trained to detect and analyse even minute changes in facial expressions and skin pores.

According to his claims, the software creates a pie chart, with the red segment representing a negative or anxious state of mind.

He claimed the software was intended for “pre-judgement without any credible evidence”.

The Chinese embassy in London did not respond to questions about the use of emotional recognition software in the province but said: “The political, economic, and social rights and freedom of religious belief in all ethnic groups in Xinjiang are fully guaranteed.

“People live in harmony regardless of their ethnic backgrounds and enjoy a stable and peaceful life with no restriction to personal freedom.”

The evidence was shown to Sophie Richardson, China director of Human Rights Watch.

“It is shocking material. It’s not just that people are being reduced to a pie chart, it’s people who are in highly coercive circumstances, under enormous pressure, being understandably nervous and that’s taken as an indication of guilt, and I think, that’s deeply problematic.”

Suspicious behaviour

According to Darren Byler, from the University of Colorado, Uyghurs routinely have to provide DNA samples to local officials, undergo digital scans and most have to download a government phone app, which gathers data including contact lists and text messages.

“Uyghur life is now about generating data,” he said.

“Everyone knows that the smartphone is something you have to carry with you, and if you don’t carry it you can be detained, they know that you’re being tracked by it. And they feel like there’s no escape,” he said.

Most of the data is fed into a computer system called the Integrated Joint Operations Platform, which Human Rights Watch claims flags up supposedly suspicious behaviour.

“The system is gathering information about dozens of different kinds of perfectly legal behaviours including things like whether people were going out the back door instead of the front door, whether they were putting gas in a car that didn’t belong to them,” said Ms Richardson.

“Authorities now place QR codes outside the doors of people’s homes so that they can easily know who’s supposed to be there and who’s not.”


There has long been debate about how closely tied Chinese technology firms are to the state. US-based research group IPVM claims to have uncovered evidence in patents filed by such companies that suggest facial recognition products were specifically designed to identify Uyghur people.

A patent filed in July 2018 by Huawei and the China Academy of Sciences describes a face recognition product that is capable of identifying people on the basis of their ethnicity.

Huawei said in response that it did “not condone the use of technology to discriminate or oppress members of any community” and that it was “independent of government” wherever it operated.

The group has also found a document which appears to suggest the firm was developing technology for a so-called One Person, One File system.

“For each person the government would store their personal information, their political activities, relationships… anything that might give you insight into how that person would behave and what kind of a threat they might pose,” said IPVM’s Conor Healy.

“It makes any kind of dissidence potentially impossible and creates true predictability for the government in the behaviour of their citizens. I don’t think that [George] Orwell would ever have imagined that a government could be capable of this kind of analysis.”

Huawei did not specifically address questions about its involvement in developing technology for the One Person, One File system but said: “Huawei opposes discrimination of all types, including the use of technology to carry out ethnic discrimination. 

“As a privately-held company, Huawei is independent of government wherever we operate. We do not condone the use of technology to discriminate against or oppress members of any community.”

The Chinese embassy in London said it had “no knowledge” of these programmes.

IPVM also claimed to have found marketing material from Chinese firm Hikvision advertising a Uyghur-detecting AI camera, and a patent for software developed by Dahua, another tech giant, which could also identify Uyghurs.

Dahua said its patent referred to all 56 recognised ethnicities in China and did not deliberately target any one of them.

It added that it provided “products and services that aim to help keep people safe” and complied “with the laws and regulations of every market” in which it operates, including the UK.

Hikvision said the details on its website were incorrect and “uploaded online without appropriate review”, adding that it did not sell or have in its product range “a minority recognition function or analytics technology”.

Dr Lan Xue, chairman of China’s National committee on AI governance, said he was not aware of the patents.

“Outside China there are a lot of those sorts of charges. Many are not accurate and not true,” he told the BBC.

“I think that the Xinjiang local government had the responsibility to really protect the Xinjiang people… if technology is used in those contexts, that’s quite understandable,” he said.

The UK’s Chinese embassy had a more robust defence, telling the BBC: “There is no so-called facial recognition technology featuring Uyghur analytics whatsoever.”

Daily surveillance

China is estimated to be home to half of the world’s almost 800 million surveillance cameras.

It also has a large number of smart cities, such as Chongqing, where AI is built into the foundations of the urban environment.

Chongqing-based investigative journalist Hu Liu told Panorama of his own experience: “Once you leave home and step into the lift, you are captured by a camera. There are cameras everywhere.”

“When I leave home to go somewhere, I call a taxi, the taxi company uploads the data to the government. I may then go to a cafe to meet a few friends and the authorities know my location through the camera in the cafe.

“There have been occasions when I have met some friends and soon after someone from the government contacts me. They warned me, ‘Don’t see that person, don’t do this and that.’

“With artificial intelligence we have nowhere to hide,” he said.


What good data mean for black youth in foster care – The Globe and Mail

Without this kind of data, can’t identify issues and possible policy approaches and solutions. One of the better elements of the former Liberal government was the emphasis its anti-racism strategy placed on better data (unlikely to survive under the current government, however):

Not enough snacks, not enough privacy, not enough allowance: Many of the complaints shared by the teenagers at the Power Up conference in Mississauga were common, even timeless.

But other concerns voiced by many of the 130 attendees, all black and all current or former Ontario foster youth, were less universal.

Some spoke about not being invited out to restaurant dinners with their foster families, or not being trusted with a house key. One 19-year-old young man started crying at the microphone. He had bumped into his sister at the conference, but they’ve been in different foster homes for so long that he didn’t recognize her.

After the formal workshops and panel discussions, kids hugged, took selfies and made plans to hang out again. “This cannot end here,” one girl declared loudly. It was a sweet, important gathering and it wouldn’t have happened without race-based data.

“The issue of the African-Canadian community and how it experiences child welfare is something that the community has been speaking to for decades,” said Kike Ojo, the program manager for One Vision, One Voice, an organization dedicated to improving the lives of black youth in foster care.

For years, she said, black parents had been sharing their anecdotal experiences of children’s aid services across Ontario: of being watched more closely than white parents; of having their children apprehended at higher rates; and of having black youth placed largely with non-black foster families, which might be loving but were often unable to help them cope with the daily realities of racism.

But it wasn’t until 2015, when the Toronto Children’s Aid Society released race-based data of the children in its care, that those who work in child welfare began taking the issue seriously. That year, 30 per cent of children in Toronto foster care were black, though only 8.5 per cent of the city identifies that way.

“That was the first time that I saw the data in print,” Ms. Ojo said in an interview. “I can’t overstate how important Toronto doing that was to getting to this moment.”

One Vision, One Voice was formed within months of the data release. It’s since toured cities around the province to give presentations on anti-black racism and equity practices, as well as to hear the experiences of youth, care workers and both biological and foster parents.

It’s also made a list of 11 recommendations to the Ontario Association of Children’s Aid Societies (OACAS), which cover everything from recruiting more black foster parents, especially those who can care for babies, to providing better transition services for youth approaching the age of 21 to help them avoid poverty and homelessness as adults.

And it’s recruiting both youth and black care workers for advisory councils, meant to be a permanent resource for OACAS to draw on when needed.

“So often the system could use a resource to help them … understand how a particular policy or program could impact black youth,” Ms. Ojo said. “Today, other than tapping a random black youth on the shoulder, they don’t have somewhere to go to kind of get that support.”

The Toronto CAS has committed to collecting and releasing race-based statistics annually: In 2017, 34 per cent of children in care were black, compared with 9 per cent of the city’s population, a small but disappointing increase that shows that identifying a problem is just the first step in solving it.

Elsewhere in the province, the current state of such data is patchwork and full of holes. In April, an Ontario Human Rights Commission (OHRC) report on black and Indigenous youth in care stated that a full 20 per cent of the province’s 49 children’s aid societies aren’t collecting it. Perhaps worse, some that do use cringe-worthy, outdated terminology: More than one tracks “mulatto” or “Gypsy” children.

In December, 2017, the Liberals’ Ministry of Child and Youth Services made collecting race-based data on foster children mandatory. All children’s aids societies were supposed to start entering information into an online portal this past February, but a number still don’t have access to the tool.

And while analysis is meant to start in early 2019, there’s been an election in the interim, and the new Conservative Ministry of Children, Community and Social Services hasn’t yet made any announcements about foster care.

That’s a tenuous state for information that’s crucial to providing Ontario’s 12,000-plus foster children with services that allow them to become healthy adults with a strong sense of self. A sense of belonging was another common desire shared by the youth at Power Up, the first gathering solely for black Ontario foster youth.

One 15-year-old said she had been to 15 different schools. She had also never had a black foster parent. “This is the family I’ve been searching for, for a really long time,” she said. A hopeful moment, thanks to good data.

via What good data mean for black youth in foster care – The Globe and Mail

Ontario children’s minister seeks racial data on kids in care

Always good to have better data to ensure improved analysis:

Ontario’s children’s minister says he will direct the province’s 47 children’s aid societies to collect race-based data as part of an effort to reduce the high number of black kids in care.

“I believe in data collection,” Michael Coteau told a conference on Thursday marking the beginning of a province-wide push to change the way children’s aid societies interact with black families.

“It is my intention in the very, very near future to mandate all children’s aid societies to collect race-based, disaggregated data,” he told child welfare officials and black community leaders at the gathering.

Black community leaders have long called for the collection of race-based data, arguing that tackling the overrepresentation of black kids in foster and group homes begins with knowing the extent of the problem. Coteau, who was appointed children’s minister in June, has for the first time committed the government to do so.

The conference was held to discuss a report calling for sweeping anti-racism reforms. It demands that every aspect of child protection in Ontario be transformed by “anti-black racism” structures and practices.

The two-volume report, called “One Vision One Voice: Changing the child welfare system to better serve African Canadians,” was written by a committee of the Ontario Association of Children’s Aid Societies (OACAS) and funded by the Ministry of Children and Youth Services.

It was triggered by an ongoing Star investigation, which revealed that 42 per cent of children in the care of the Children’s Aid Society of Toronto were black, in a city where only 8 per cent of children are black.

Coteau reminded the conference that in 2004 he was part of a team of trustees behind the Toronto District School Board’s decision to collect race-based data on students.

“If you have no data, there’s no problem and there’s no solution,” Coteau said, citing the board’s superintendent of education at the time.

Coteau, also responsible for anti-racism in Ontario, promised that his ministry would go much further than data collection.

“Substantial reform is on the horizon,” he said, referring to major changes expected in Ontario’s privately run child protection system.

He said the Liberal government would soon amend the Child and Family Services Act to modernize children’s aid societies, which last year received $1.5 billion in provincial funding. He promised to make them more accountable and transparent, and to improve the patchwork of services and care they now provide.

“We want high-quality services that reflect Ontario’s diversity, consistently delivered across the province,” he said.

After his speech, Coteau refused to answer questions from the Star. He would not say whether his ministry was prepared to fund recommendations in the report calling for anti-racism reforms.

The OACAS, which represents all but four of the province’s societies, fully backs the recommendations in the anti-racism report, says Mary Ballantyne, the association’s CEO. She stressed the reforms can only be implemented with extra provincial funding.

The call for implementation funds was echoed by Ontario’s human rights commissioner, Renu Mandhane, and the province’s Child Advocate, Irwin Elman.

“We will continue to push the minister,” Elman said in his speech to the conference, adding he told Coteau, “You can’t just walk away now from this report. You need to provide resources.”

Source: Ontario children’s minister seeks racial data on kids in care | Toronto Star

CAS study reveals stark racial disparities for blacks, aboriginals

The disparities are quite striking and like all studies, force questioning into the reasons why, including implicit bias:

New research that for the first time calculates disparity in Ontario’s child protection system has found that aboriginal and black kids are far more likely to be investigated and taken into care than white children.

The figures are especially stunning for aboriginal children. They are 130 per cent more likely to be investigated as possible victims of child abuse or neglect than white children, and 15 per cent more likely to have maltreatment confirmed.

Aboriginal children are also 168 per cent more likely to be taken from their homes and placed into care.

The huge disparity is “symptomatic of the system that’s failing our kids,” says Steven Vanloffeld, executive director of the Association of Native Child and Family Service Agencies of Ontario.

The study also found that black children are 40 per cent more likely to be investigated for abuse or neglect than white children, and 18 per cent more likely to have maltreatment confirmed. But the likelihood of going into care is lower. Black children are 13 per cent more likely to be taken from their homes and placed with foster parents or in group homes.

Margaret Parsons, executive director of the African Canadian Legal Clinic, blames the disparity on the “harsher lens” children’s aid societies use when investigating black families.

“What they might not consider abuse or neglect within a white or non-African Canadian family, they will consider abuse or neglect in one of our families,” she says. “This is not a matter of erring on the side of caution. We feel it is punitive.”

The provincial government, which regulates the child protection system, must make the development of an African Canadian child welfare strategy a priority, she adds.

The estimates were extracted from the government-funded Ontario Incidence Study of Reported Child Abuse and Neglect, compiled in 2013. A team of researchers, led by University of Toronto Prof. Barbara Fallon, examined a representative sample of 4,961 child protection investigations conducted by 17 children’s aid societies. The cases involved children up to age 14.

Of the dozen specific ethnic and racial categories examined, only black and aboriginal children were taken into care at rates higher than white kids.

The study was presented to more than 70 senior children’s aid society officials at a June 7 meeting in Toronto.

The disparity study calculated the relative likelihood of certain groups being involved with the child protection system. It differs from the study on disproportionate representation revealed by an ongoing Star investigation, which found that on a September day in 2013, 42 per cent of kids in the care of the Children’s Aid Society of Toronto had at least one parent who is black. Only 8 per cent of the city’s under-18 population is black.

The disparity results coincide with mounting outrage about the disproportionate number of aboriginal and black children in care. Parents and leaders in these communities have for years blamed discrimination and a lack of services for struggling families.

Source: CAS study reveals stark racial disparities for blacks, aboriginals | Toronto Star

Ontario children’s aid societies agree to collect race data


But data should be used not only to identify bias and ensure consistent treatment but also by communities to discuss aspects that may be internal:

Children’s aid societies in Ontario have agreed to collect data on the race of children and families they serve, a move that comes after mounting outrage about the high number of black and aboriginal kids in care.

Mary Ballantyne, CEO of the Ontario Association of Children’s Aid Societies, says societies will have a “consistent approach” in place within a year to gather the race-based data.

“We need to have this data,” Ballantyne said. “It will help us figure out what kinds of services need to be provided to what kinds of kids.”

The goal is to ensure that families and children served across the province are being treated equitably, she added. “It is a top priority.”

Ballantyne vowed to make the province-wide race data public as soon as it is collected and analyzed.

The societies are making the commitment after an ongoing Star investigation found that 42 per cent of children and youth in the care of the Children’s Aid Society of Toronto have at least one parent who is black. Only 8 per cent of the city’s under-18 population is black.

Black community leaders have complained for years that their children are taken into care in disproportionate numbers. They’ve been pushing the government to order the collection of race-based data, arguing that the child protection system is biased against black families.

Everton Gordon, interim CEO of the Jamaican Canadian Association, called the societies’ commitment to collect race-based data a step in the right direction. He insists, however, that the statistics be used to trigger programs and reforms that will reduce the number of black kids in care.

“It’s more than just a numbers game,” Gordon said in an interview. “It speaks to a system that has embedded biases that produce disproportionality for different marginalized groups.”

Source: Ontario children’s aid societies agree to collect race data | Toronto Star

Black kids stay longest in care, CAS study shows

Black_kids_stay_longest_in_care__CAS_study_shows___Toronto_StarMore data on children’s aid in Toronto (see earlier article):

Black children in Toronto stay longer in foster care and group homes than any other group of kids.

A survey conducted by the Children’s Aid Society of Toronto found 45 per cent of black children taken from parents in the 2008 fiscal year spent more than 12 months in care.

Only 20 per cent of white children taken during that period spent more than a year in care. For children with Asian parents, the number was 18 per cent.

The study looked only at families who came into contact with children’s aid for the first time. Of those families, 126 of them had children taken into care.

The numbers were part of a Toronto society analysis that also highlights what black parents have been angry about for years: their children are taken into care at rates far higher than white children.

It confirms numbers first reported by the Star in December 2014 — 31 per cent of children in the society’s care are black and a further 11 per cent had one parent who is black. In Toronto, 8.2 per cent of people under 18 are black.

Black kids stay longest in care, CAS study shows | Toronto Star.

Ontario may collect race-based data on kids in care

Always controversial to collect race-based data but without data, hard to know what is happening and what measures could be taken to address problem areas:

The probe revealed for the first time that 41 per cent of children and youth in the care of the Children’s Aid Society of Toronto are black, though only 8.2 per cent of the city’s under-18 population is black.

MacCharles, who became Minister of Children and Youth Services last summer, is so concerned by those numbers she’s considering a province-wide count of black kids in care to determine the extent of the challenge. Few of Ontario’s 46 children’s aid societies track such data, and those that do keep the statistics secret.

Black community leaders have complained for years that their children are taken into care at rates far higher than white children. They say it is hard to get government to pay attention without hard statistics.

“I think there’s a lot more receptivity to looking at (race-based data) in this sector and beyond,” MacCharles said of the government’s current attitude. “We’re also looking at this notion of disaggregated data, which includes black children and youth in care, in schools, and in our youth justice system,” she added.

Without committing to making such data public, MacCharles told the Star: “My bottom line is, any data that helps improve the security and safety of children, I’m willing to have a hard look at.”

Ontario may collect race-based data on kids in care | Toronto Star.