Digitalizing racial terror in Salvador/Brazil: Facial recognition use by police and the update of historical patterns of state violence against Black communities

Facial recognition as an issue in Salvador, Brazil…

‍

In 2023, there was news of a Black man being wrongly arrested by police in June for 26 days because of a false positive by the facial recognition used by Bahia Public Security Secretariat. That was not the first time such a thing. In 2020, a young man was wrongly approached by police officers and put at gunpoint in front of his mother.

‍

The state of Bahia and its capital Salvador started using facial recognition for police surveillance in December 2018. Connecting this biometric technology with the CCTV system, this system arrested between 2018 and 2023 more than 1000 people - as said by official records. Its implementation was defined by a non-transparency and in no compliance with Brazilian legal and constitutional guidelines and it was invested more than R$ 680 millions - almost 140 millions US Dollars.

‍

Initially the system was used in big events like New Year's Eve, Carnaval and local holidays in the city of Salvador. Then, it started being used in points of high circulation of people like the subway system, soccer stadium, tourist landmarks and other public spaces. Then in 2021 it initiated a process of expansion of the system beyond the initial 100 cameras in Salvador and implemented cameras in more than 70 cities in the interior of Bahia.

‍

Brazil and the historical patterns of racial violence and surveillance against Black people

‍

****Talking about surveillance in Brazil asks us to speak about slavery and a penal system deeply rooted in anti-black racism. Professor Ana Luiza Flauzinaargues that the architecture of punishment in Brazil has roots on slavery , class control of enslaved people and in creating a legal system against those groups. These patterns continue in the penal system, even with slavery abolition, with Black communities being persecuted by the police.

‍

Sociologist Vilma Reis adds to this argument by pointing out that Bahia state police - and especially in Salvador - was founded to promote  counter-insurrection against quilombos (social forms of resistance by Black people during slavery ) and other kinds of revolts against slavery. This institutional memory continues after abolition and is reinforced by biology based criminal anthropology which tries to establish the idea of Black (and indigenous) people's tendency to crime.

‍

Current Brazilian penal system is defined by mass incarceration and police violence against Black people and our communities. With a criminal policy defined by capital protection  and the “War on Drugs” framework - a logic that as Michelle Alexander already established is connected with targeting of Black communities - we have a incarceration mainly composed by non-white and lower income people.

‍

In a recent study by the Rede Observatórios de Segurança - Network of Security Observatories -, it was found that 94,76% of deaths result of State intervations were of Black people. Bahia has a proportion of Black population of 80,80% which express how racialized this State violence is.

‍

Facial recognition issue through the lens of race

‍

****Facial recognition and its use for surveillance has been widely criticized because of their potential to promote harm and violate several fundamental rights and public liberties.  Besides the worries with data protection, privacy and right to assembly, facial recognition is also a technology with a high tendency to promote and increase racial and gender discrimination.

‍

Joy Buolowaimi, Ruha Benjamin, Simone Browne, Timmit Grebhru are some of the scholars who pointed to racial and gender bias in computer vision systems, especially facial recognition. It has been shown that these systems disproportionately mistake non-white faces, especially Black women.

‍

But what I worry the most is not only the algorithmic injustices that come with technical issues of those systems, but how this new tech is added into already exclusion and violence based social structures, especially policing.

‍

Post-slavery contexts where police became an expression of racism as a power structure - like Brazil and the United States - means police archives and criminal data defined by these practices of targeting racialized and impoverished communities. That means any technology that depends on previously collected data to work will end up reproducing the patterns of this historical collection. The past becomes the future, but updated.

‍

Updating racial terror through digital technology

‍

In this sense I see facial recognition implementation in Bahia and in the city of Salvador as an update of historical patterns of persecution and violence against Black people. It helps the State open its capacity of monitoring and promoting violence into the territories, controlling flows of people widely, and producing racial differences through digital technology.

‍

A report done by Rede Observatórios de Segurança (Network of Security Observatories)  in 2019 showed that 90% of the arrests done in Brazil were of Black people.  Half of these arrests were in Bahia

‍

In my own research, I identified 408 arrests between December 2018 and August 2022 in Bahia, using reports done by the Public Security Secretariat. In 145 of those, there is no information about the person arrested and the accusation, showing the lack of transparency in the handling of public security affairs.

‍

Taking a look at the accusation that justified the arrest in the ones where there is more information allowed to use with criminal offenses end up more targeted by this technology.

‍

You can see that “Robbery” and “Drug offenses” being the champions of arrests in the use of facial recognition in Bahia, expressing the patterns of capital control and “War on Drugs”. This follows the patterns of incarceration data we have in Brazil that shows “Crimes against the patrimony” and “Crimes against the Drug Law '' as the primary reasons for arrests.

‍

Both criminalization choices tend to deeply affect Black and impoverished communities in Brazil and materialize how facial recognition tends to repeat already established patterns of State repression.  Also, following what Jackie Wang defines as “algorithmic policing”, this tech helps police create a discourse of modernity and neutrality through scientific truth. In this sense, facial recognition not only updates racial terror and repression in its material work but also in the police discourse.

‍

There is also the issue of lack of transparency between police and the public about the use of facial recognition. Asked about false-positive rates, the Public Security Secretariat responded saying there was none. But news of wrongly arrested people because of the errors from the biometric tool continue to go by - and also, how many more are of these arrests that didn’t go to the public?

‍

Salvador, also called the Black Rome because of its majority Black population and importance to afrobrazillian religions, was the initial laboratory for the use of this technology and where this update was fully tested and defined by public authorities as a case of success. In a city that expresses Black culture, political power and life, tech comes to express an update of the hard truth of Brazil racial defined State violence.

‍