Some US states stop law enforcement facial recognition systems

Between Europe and the United States, facial recognition technologies continue to divide and worry. The recent proposal for a Regulation on European Artificial Intelligence includes among the “high-risk” uses all the applications of biometric identification systems in places accessible to the public, with the purpose of social scoring or the suppression of crimes, reaffirming the prohibition of use for of mass surveillance. In the United States, the problem is perceived with even greater force, especially with regard to racial discrimination that can result from the application of facial recognition software. According to a report by Amnesty International, one of the biggest problems is that these systems can be used without the subject’s knowledge, and precisely this lack of transparency could lead people to be more fearful and cautious in expressing their dissent. . An even greater danger concerns people of color. Facial recognition systems have historically proven to be less accurate towards non-Caucasian ethnicities, because the algorithms are mostly created from datasets consisting predominantly of images of white men. A 2018 MIT publication demonstrated the ineffectiveness of Microsoft and IBM systems in correctly identifying dark-skinned individuals, and other studies found flaws in Amazon’s Rekognition system, which is inaccurate in recognizing female faces and people of color. Although the companies have partially refuted the methods used by the researchers, all three, in June last year, imposed a self-imposed regulation with restrictions on the sale of these technologies to the police, at least until a law is in place. national on the matter. Both Amazon and Microsoft have specified that any law must have its basis in the most rigorous respect for human rights, and IBM has gone even further, giving up the sector in its entirety, denouncing, in fact, its incompatibility with an application without distortions. . But after a year of reflections, triggered by the killing of George Floyd and the protests of the Black Lives Matter movement, which led civil rights advocates to ask for greater certainty in the relations between law enforcement and citizens, especially when the racial prejudice, everything is still at the starting point. The impetus given by the moratoriums of large companies was of little use. At the national level, there is only a proposal by the Democrats to pause the use of these technologies until a coherent legislative framework is reached, while the only initiatives on the subject have been taken in the few states more attentive to respect for rights. Also in the wake of protests, New York State passed a law requiring police departments to be transparent about the use of surveillance systems, revealing how facial recognition technologies have been used in over 22,000 cases in the past. three years, also to trace the protesters of the Black Lives Matter movement. Also in Washington state (as well as in Maryland and Virginia) similar technologies have been used to identify protesters accused of crimes, which led to the announcement, a few weeks ago, of the blocking by July of the facial recognition system in use by the law enforcement in Washington and Virginia.