The EU shields its borders with facial recognition systems only for non-EU | Technology

The European Union finalizes the implementation of facial recognition systems at all its borders. They will be used to collect the biometric data of the so-called citizens of third countries (those from outside the Schengen area), who until now only had their fingerprints taken, and compare them in real time with the lists of people indicated by both the authorities local as well as community. The information will be stored in a large database that will be managed by the European Commission and will be accessible to the Member States. The application of this technology at the borders, which the EU wants to have operational before the end of the year, has sparked complaints among activists, politicians and technologists who consider that it can discriminate against and criminalize irregular immigrants.

Facial recognition systems allow people to be unequivocally identified from an image of their face. Each human face has a particular arrangement of facial features, which does not vary with aging; With the help of artificial intelligence algorithms, it is possible to extract a specific pattern, a kind of unique identifier, and detect it in photographs or videos. It is an identification method as reliable as fingerprints. But while this last system requires the collaboration of the subject, who must place their fingers on a reader, facial recognition only needs to feed on digitized images. For example, footage from a street surveillance camera.

The systematic use of this technology is contemplated in the Entry and Exit System (Entry/Exit System, EES), an EU project approved in 2017 and endowed with a budget of around 650 million euros. Its objective is to automate and digitize the registration of biometric data of citizens of third countries when they enter or leave through the community borders. The EES does not apply to European citizens, of whom Brussels does not store their data on their faces, but it will do so with those who come from outside the 27.

Until now, facial recognition was only used in Europe in criminal investigations, mainly terrorism. The draft European Artificial Intelligence Regulation, which is expected to come into force next year, classifies facial recognition as a “high-risk technology”, which means that it is only used for “crime prevention, detention or investigation purposes”. serious or terrorism”. The small print of the article, however, allows it to be used at the borders.

The EU’s caution with facial recognition responds to the fact that it is considered a mass surveillance technology (people can be identified without their consent) and that, far from being perfect, the algorithms fail. Various studies prove that the accuracy of these systems declines if they are used with groups that are not Caucasian men: they fail more with women, young people and racialized people. In the US, the case of Robert Williams, a black person who was arrested when the algorithm mistook him for another, sparked a movement that is causing police departments in many cities across the country to stop using this technology.

Why collect more biometric data

Brussels’ arguments are clear: the Entry and Exit System will make visitor traffic more efficient. “Each Member State is responsible for setting up its own platform for collecting and processing this biometric data, as well as for providing adequate security measures for it,” says a spokeswoman for the European Commission. The same sources indicate that these data, which will be physically housed in the Belgian capital (on eu-LISA servers), should be stored only for as long as they are necessary, although they do not specify whether to speak of months, years or decades.

One of the bodies that must ensure that there are no excesses in the handling of data as sensitive as these is the European Data Protection Supervisor (EDPS), an independent EU office that has been very critical of the Project. Already in 2015 he showed his reservations about it, which he confirmed last year by requesting the prohibition without exceptions of facial recognition in public spaces. As EL PAÍS has learned, the Supervisor’s position has not changed since then.

The fear of critics with the project of digital borders (EES) is that it harms a group, that of irregular immigrants, already vulnerable in itself. Because facial recognition systems will also be applied to those who arrive in Europe by boat. “The discrimination posed by the use of this technology at borders has two aspects,” says Ana Valdivia, a doctor in computing and artificial intelligence and a postdoctoral researcher at King’s College London. ”The first is the technique: the algorithm can make mistakes, misidentify. These tools have biases, they do not work equally with all skin colors, genders and ages. There is very little digital data of minors, for example, that is why they fail so much with them, ”she illustrates.

“The second perspective is political: how technology is used,” continues Valdivia, who has studied in depth the application of artificial intelligence at borders. She was one of the promoters of an open letter addressed to the Government in January and signed by fifty experts and associations, in which control mechanisms were requested to review the algorithmic procedures, so that a failure of the tool could not leave someone without a visa, or the incorporation of human rights observers at the borders, among others.

“The shadows of this project are in the criminalization of migration,” says Ismael Cortés, deputy for United We Can. In March, Cortés defended a Non-Law Proposal (PNL) on the collection of biometric data at the border in the Interior Commission of Congress, in which the Government was urged to guarantee that “artificial intelligence technologies aimed at facial recognition at the border” do not generate “discriminatory biases” or “involve a risk for the exercise of civil and political liberties”. The PNL, promoted with the intention of opening this debate, was approved with the support of the PSOE, the other part of the Spanish Government.

“The EU is going to put two types of subjects in the same bag: those who until now were the object of biometric data collection, that is, those suspected of terrorism or organized crime, and those with an irregular residence situation in a country of the Schengen area”, criticizes the En Comú Podem deputy. That is why it is key to know if the databases will be mixed and what measures are going to be taken to detect false positives (when the algorithm fails and links the newcomer to Europe with one of the suspected criminals already registered).

The simple fact of storing information as sensitive as facial data, Valdivia recalls, always involves risks: it can be stolen or fall into the wrong hands. In Afghanistan, when the Taliban regained control of the country, they used databases with the irises, facial features and fingerprints of former collaborators of the regime they overthrew to persecute them, according to various media reports. “Imagine that Hungary decides to use its database to identify who is homosexual and who is not,” slides the researcher.

You can follow THE COUNTRY TECHNOLOGY in Facebook Y Twitter or sign up here to receive our weekly newsletter.