In France, facial recognition is gaining ground and causing concern

This is one of the campaign themes of these regional: facial recognition. Among the candidates, Valérie Pécresse, the president of the Ile-de-France region and Laurent Wauquiez, president of the Auvergne-Rhône-Alpe region declared themselves in favor of it. With the imprint of your face, this technique allows you to unlock your smartphone or go through security gates at the airport. But it can also identify terrorists on file wanted in public transport or on public roads. This last use is prohibited in France so far.

The law does not prohibit this tool in itself, but it very strictly regulates its use. In particular, it requires obtaining the consent of the person, with a few exceptions. Patrice Navarro, is a lawyer specializing in data protection, he advises companies on this subject. “Everyone has been wary of facial recognition for a long time, he explains. This is a subject that is watched very closely by the CNIL, which continues to follow experiments. It has thus refused the use of facial recognition in high schools or to prevent hooligans from entering stadiums. France is not the most restrictive country, but it is one of the countries that thinks about it the most on the other hand.

An experiment in the Paris metro

France has even already tested several experiments that approach facial recognition. Latest: that of Cannes during confinement, or just after, in the Paris metro at Châtelet-les-Halles, to check that people were wearing their masks. It is the start-up Datakalab which has set up this device. Yet Xavier Fischer, its managing director, affirms it: it was not a question of identifying the travellers. “We quantified the mask-wearing rate, he says. And we saw, for example, that the mask was worn better during the week than on weekends, in the morning than in the afternoon. All this to make distributions masks, sound announcements. Valérie Pécresse used this data to communicate with the general public, never to verbalize. The data was only delivered 20 minutes later in order to be certain that the person was far away, and that it was not not thanks to the technology that we were verbalizing."

Last June, the CNIL had sounded the alarm and the experiment had been interrupted because of this question of consent. But Xavier Fischer will be able to continue the experiment in the Paris metro, thanks to a decree valid for one year, recently published by the Minister of Transport. "The CNIL told us: 'The only way for you to legally resume your treatment is to rely on a decree issued by the Ministry of Transport.' Valérie Pécresse's office worked with the RATP and the ministry, and this famous decree came out on March 11 to allow for one year the analysis of the image for statistical purposes in the transport networks.

The "foot-in-the-door strategy"

This is what makes defenders of individual freedoms say that this technology is advancing hidden and gaining ground in France. It is the position of La Quadrature du Net which alerts: according to them, the situation in France is serious. Martin Drago is a lawyer within the association, he closely monitors the progress of this technology on the territory. "It's a bit of a foot-in-the-door strategy, he analyzes. For twenty years, we have deployed massive video surveillance cameras in France. And little by little, we are starting to impose ourselves and deploy a layer additional algorithmic: we detect masks, suspicious behavior, etc. And then tomorrow, maybe everyone will have to be identified and identifiable in the street when we walk. we're all a bit suspicious."

The issue of databases

The challenge for these regional elections is the deployment of this technology during the 2023 Rugby World Cup and the 2024 Olympics in Paris. Because the government and French companies want to make these two major events a laboratory and a showcase for new security technologies. They would make it easier to enter the stadiums and identify undesirables in the fanzones.

But on this issue, there are still obstacles. “Technically, we can have false positives, which is absolutely to be avoided, advances Maître Patrice Navarro, who is working on this file. Depending on how artificial intelligence has been trained, this can lead it to make mistakes. connection to public security databases. And we all know it: the more databases of this type are assembled, the more Big Brother-style behavior is to be feared. It remains very difficult to set up such a system today with enough guarantees to ensure that we don't go too far..."

At the same time, a European regulation on artificial intelligence has been in preparation in Parliament since April. But its application should take several more years.