The European Parliament adopted, on Wednesday June 14, the AI Act (for Artificial Intelligence Act), a text regulating artificial intelligence. The idea is to offer “a regulatory framework for placing on the market with the triptych safety, health, fundamental rightssummarizes Arnaud Latil, lecturer in public law at Sorbonne University. The European Union [UE]as it did before with the General Data Protection Regulation, uses with the AI Act the so-called “risk-based approach” doctrine”.
As it stands, the regulation adopted by Parliament classifies artificial intelligence (AI) systems according to a level of risk ranging from “minimal” to “unacceptable”. Prohibitions are rare there: they concern uses contrary to European values, such as the “social credit systems” or mass video surveillance used in China. MEPs also want to remove the exemptions provided for by the European Commission to authorize remote facial recognition of people in public places by law enforcement. When it comes to “high risk” AI systems, they will have to comply with the strictest transparency, risk management and data governance regime.
Non-governmental organizations (NGOs) and civil society remain reserved, however, judging that for Europe’s migration policy, an area in which AI is already used at EU borders, the regulation does not give enough guarantees.
The blind spot of migration policies
“The AI Act is the first regulatory law on a regional scale, and when we see the growing development of artificial intelligence, it’s a good thing. But it is clear that it does not go far enough, that the devil is in the details », regrets Petra Molnar, lawyer and associate director of the Refugee Law laboratory at York University (Canada). Here, the detail is Schedule 9 of the AI Act, by which the text waives regulation of AIs belonging to large-scale computer systems if these are already used in the context of migration management.
Thus, the sBMS, developed by the French companies Idemia and Sopra Steria, aggregates the fingerprints and portraits of more than 400 million third-country nationals. However, it is precisely these databases on which software under development since September 2020 intends to rely: ITFlows, funded by the EU to the tune of 4.8 million euros. This uses an AI to predict migratory flows, thus identifying the “risks of tension” linked to the arrival of migrants at EU borders.
You have 54.65% of this article left to read. The following is for subscribers only.