4
Oct
2021
Watch
Artificial intelligence in criminal law and its use by the police and judicial authorities in criminal matters (debate)
Mr President, ladies and gentlemen, while Parliament is preparing to examine the proposal for a regulation on artificial intelligence, with this report we are giving a clear message and already making an important point: In our view, there is no place in Europe for mass biometric surveillance and security and the fight against crime cannot be at the expense of citizens' fundamental rights. Indeed, identification by means of biometric data in publicly accessible places risks leading to serious abuses of the right to privacy and other underlying principles of our democratic systems. The European Data Protection Supervisor said: these systems would have a direct negative impact on the exercise of freedom of expression, assembly, association and even freedom of movement. Let us then think about what could happen in places that are not so attentive to the separation of powers or fundamental freedoms, be they states or cities. The risk of abuse is too high. That is why we believe that the exceptions to Article 5 on prohibited practices in the proposal for a regulation should be removed. Similarly, predictive techniques for law enforcement bring with them a very serious risk of discrimination, in addition to the lack of evidence on their accuracy, undermining one of the fundamental foundations of our democratic systems, namely the presumption of innocence. No human oversight or error-free datasets will be sufficient to ensure that such decisions by AI systems are made respecting constitutional guarantees and fundamental rights of the Union, even where these decision-making processes are reversible. A fortiori, these systems cannot undergo a mere conformity self-assessment before being placed on the market, as proposed by the Regulation in its first draft before us. A self-assessment exposes people to unacceptable risks of errors and violations, which would only be discovered later by the supervisory authorities if they had the means to do so, and this would result in damage that has now occurred, even irreparable, to people's lives. We already have the world's most advanced data protection laws in the EU today. For us it is a model, a model that we want to bring to the rest of the world, and we cannot afford to retreat even a millimeter from this setting when we find ourselves regulating artificial intelligence. In this area too, we must fully protect the rights of citizens. I think this is how we can act for a Europe that has its own model centered on the human rights of artificial intelligence.