Article | AI with Rights | Digital Platforms and Markets

Banishment, moratorium, regulation: movements surrounding facial recognition

by Gabriela Vergili, Iasmine Favaro, Mariana Rielli and Rafael Zanatta

In the article titled Regulating AI and Machine Learning: Setting the Regulatory Agenda, Julia Black and Andrew Murray argue that the way in which society receives and absorbs the potential of new technologies is largely determined by the regulatory and governance models that are applied to them. Among the technologies currently at the center of the regulatory debate, facial recognition stands out. Although this is not a new debate (in 2009, the Madrid Declaration already called for a moratorium on facial recognition for surveillance purposes), recently, the advancement of the technology has led to movements for its regulation and, in some cases, for its partial or complete ban.

This is not without reason. In a recent report, The Guardian revealed that Germany conducted a trial of the tool in Berlin, and due to its success, plans to implement facial recognition in 134 train stations and 14 airports. In France, there is also an accelerated movement toward the application of the technology, and in the United Kingdom, supporters of Brexit view the country’s exit from the European Union as beneficial because they want the rapid installation of this technology, among other things. This distancing means escaping existing regulations on privacy and personal data protection, as well as future laws concerning the use of facial recognition. The rapid advancement of the technology raises concerns, not only from organized civil society.

Major technology companies, such as Microsoft, do not deny the potential problems arising from the use of facial recognition, but at the same time, they do not endorse the argument that the technology should be banned. There is a trend among the business sector to pressure governments to establish regulations, parameters, and principles for the application of facial recognition, while also developing their own guidelines.

According to the article “Facial Recognition: It’s Time for Action” by Brad Smith, President of Microsoft, failing to immediately establish regulatory parameters for the use of facial recognition and banning its use means that, in the future, the use of the technology will be inevitable and will have an even more negative impact from the perspective of oppressive surveillance. From a practical standpoint, the company outlined the three main purposes of facial recognition regulation: addressing biases and discrimination, protecting people’s privacy, and safeguarding democratic freedoms and human rights.

However, Microsoft’s opinion has not convinced some U.S. states and cities. In 2019, San Francisco banned the use of facial recognition by local government agencies, and in 2020, Cambridge, the city home to the Massachusetts Institute of Technology (MIT) and Harvard, also vetoed the use of the technology. These are two representative events of how the world’s most technologically advanced cities have been dealing with the problems posed by the application of facial recognition.

Between banning and immediate implementation, there are institutions that advocate for a moratorium on the use of facial recognition technologies. More than forty groups signed a letter drafted on January 27 by the Electronic Privacy Information Center (EPIC) addressed to the U.S. Privacy and Civil Liberties Oversight Board, an executive branch agency. The argument in the letter revolves around the idea that the use of the technology should be prohibited “pending review”, because, in a recent New York Times investigation into the company Clearview AI, it was concluded that the formulation of the database and the implementation of decisions through artificial intelligence can be inaccurate and can be used to “control minority populations and limit dissent”.

The letter comes in the wake of a growing movement for a moratorium on facial recognition, which is expressed in manifestos such as the one signed by Professor Kate Crawford in August 2019, and the Declaration for a Moratorium on Facial Recognition Technology for Mass Surveillance Purposes, adopted in Tirana, Albania, in October 2019, and endorsed by around 100 organizations and over 1000 individuals. In the same month, professors Evan Selinger and Woodrow Hartzog wrote a strong essay on the subject for the New York Times, titled “What Happens When Employers Can Read Your Facial Expressions?” The arguments center around the inaccuracies and biases observed in the use of the technology, but also the dangers it presents when it works correctly. The ubiquity of facial recognition, meaning its ability to “be everywhere,” is one of the points raised. The subtitle of Selinger and Hartzog’s article summarizes the thesis: “The benefits don’t come close to outweighing the risks.

More recently, the European Union has taken a step toward adopting a moratorium on facial recognition in public spaces for a period of 3 to 5 years. The proposal emerged in a “white paper” from the European Commission, highlighting the need for risk assessment mechanisms before allowing the widespread use of facial recognition in places such as shopping malls, public squares, football stadiums, etc. However, the initiative seems not to have progressed in the more recent versions of the document, as indicated by a Reuters report.

In Brazil, similarly, the technology appears to be advancing even without any specific regulation. A study by the Igarapé Institute identified numerous cases of the technology being implemented by public authorities, with a focus on the following sectors: (i) education, (ii) transportation, (iii) border control, and (iv) public security. The private sector is no different – in 2019, the Brazilian Institute for Consumer Defense (Idec) notified the company Hering, requesting information on the use of facial recognition technology for targeted advertising purposes. Ricardo Abramovay, in his essay How to Face the Era of Total Surveillance, references the Brazilian case and the dangers posed by the advancement of the technology, in the context of initiatives such as the new Citizen’s Base Registration.

There is currently no movement in Brazil calling for a ban on facial recognition or a moratorium, but the same cannot be said for initiatives aimed at regulating it. While between 2015 and 2017, there were two legislative acts on facial recognition in the Chamber of Deputies, between 2018 and 2019, the number rose to 23, including 10 bills in progress, with various focuses: the right to information, use in protective measures, and the regulation itself.

While specific regulation on the matter has not yet emerged, it is worth noting that Brazil has a robust framework for protecting fundamental rights, including privacy, and that the General Data Protection Law (Law No. 13,709/2018), which came into effect in August 2020, also includes elements that will be important in the regulatory debate on the subject, such as the principles of necessity, quality, and non-discrimination, as well as the requirement for data protection impact assessments.