Face recognition tech gains ground in Brazil
Brazil is home to 37 countries that have adopted some form of face recognition technology. Over half of the initiatives—19—were launched between 2018 and 2019. The solutions are usually implemented in public security, transport, and border control.
A survey was conducted by the Igarapé Institute, an independent organization dedicated to the integration of the security, justice, and development agendas. The study was unveiled Thursday (Sep 19) at the 10th Seminar on the Protection of Privacy and Personal Data, organized by the Internet Management Committee of Brazil. The research covered initiatives across the country, started in 2011 and later.
Apple of discord
Authorities have advocated face recognition as a sophisticated control tool for public policies. The topic, however, has sparked controversy as it raises questions regarding privacy violation and civil control. The technology has been banned in US cities like San Francisco and Oakland, California.
However, governments, independent organizations, and even tech giants like Microsoft have stood out for the practice.
Nothing new under the sun
Researcher Louise Marie Hurel, of the Igarapé Institute, noted that implementing such an identification method is no novelty. In 2004, a bill by then Deputy Eduardo Paes, former Rio de Janeiro mayor, sought to regulate the use of face biometrics to authenticate access to tax data. Early this decade, cities started resorting to this technology.
The first projects focused on transport. Outer city buses would set recognition as a condition for users. This sector accounts for 21 projects surveyed by the study. In the last few years, the technology gained special momentum in public security, like the access to facilities monitored by CCTV. Thirteen other initiatives identified share the same purposes.
Not only governments but also companies have joined the trend. The concessionaire for one of São Paulo’s metro lines has installed cameras to gauge the mood of passengers through their face expressions so more appropriate ads could be displayed. Hering, a Brazilian clothing chain, placed similar systems in one its São Paulo stores to examine consumers’ attitudes, interests, and practices as an element to be considered when devising marketing strategies.
Legislation
The spread of this innovation has raised concern among authorities. The survey found two bills on the subject in the Brazilian Congress and 21 in city councils—eight of which in Rio de Janeiro and two in São Paulo. “Face recognition has gained silver bullet status in public security,” researcher Marie Hurel pointed out.
One of the bills under scrutiny in Congress makes face recognition in prison facilities mandatory. The other piece of legislation also stipulates the use of the technology for prison workers and even lawyers coming in and out of penitentiaries.
Risks
Rafael Mafei, professor at the University of São Paulo (USP) Law School, stressed the risks coming with this identification method. In his view, “face recognition is so dangerous when used inappropriately that it’s just not worthy of becoming popular. This could be the dream technology of authoritarian governments.”
Lawyer Bárbara Simão, of Brazil’s consumer rights institute Idec, noted that the surveys conducted by research institutes in counties like the US and the UK indicate low confidence in this technology among the public. In one of the polls, a mere seven percent agreed that this technology could be used in the selection of customized advertisement.
When deciding to implement the method, companies must take into account a number of criteria, the lawyer said. The first is whether it can really be an asset, and whether identification can be carried out by some other means. The second relates to the transparency of the operation. The third is about respecting people’s choices, rather than adopting face recognition as imposed.
How it works
Face recognition starts with the collection the image of an individual. Next, a process referred to as “normalization,” where people are classified according to patterns. Their features are transformed in reference points for analysis. These data are processed and linked to the identity of the person in question.
In authentication services, for instance, cameras produce footage while the system attempts to find a compatible face in its data base.
*The reported traveled to São Paulo at the invitation of the Internet Management Committee of Brazil and the Ponto BR Information and Coordination Center.