The Proposal for a European regulation on artificial intelligence contains, inter alia, a few statements regarding Artificial Intelligence that should not be underestimated. In essence, it is proposed that the financial supervisory authorities (and therefore also those of prudential supervision) carry out the task of supervisors of artificial intelligence (see Recital 80). The Proposal also provides that high-risk artificial intelligence systems are subjected to a conformity assessment regime (see Article 41). The Prudential Banking Supervisors are not equipped to do this and in fact it is not appropriate for them to do so. This is particularly clear at the European level. In fact, the supervisory mandate entrusted to the ECB is strict within the limits of Article 127 (6) TFEU. It is doubtful that a European regulation can entrust it with the role of supervisor of artificial intelligence without going ultra vires (we have seen in the past that the ECB has been cautious about the possibility of intervening in other matters, such as anti-money laundering). The doubt seems to be confirmed by the opinion that the ECB itself sent to the national legislators in relation to the Proposal. In reality it is desirable that the Authorities of Prudential supervision continue to do what they have always had to do: prudential supervision. If the bank uses artificial intelligence, the supervisor will have to equip itself with the necessary skills and resources (including human), but this will not make the supervisor a supervisor on the technology market, nor will it make him a guarantor of individual rights.
R Lener (Sat,) studied this question.