FESE believes that a revision of the regulatory framework is necessary, as the current legislation may have some gaps in the application of AI, taking into account the impacted sectors in order to review and/or regulate, inter alia, the responsibility that would be required for the use of AI systems. Cooperation between authorities and market participants has the potential to bring valuable outcomes. This includes a mutual understanding of benefits and risks associated with the technology and lays the ground for a wider ecosystem.
The creation of training and professional centres of excellence, along with the European Data Space (please see our response to the Commission consultation on a “European strategy for data”) are two key elements for a meaningful adoption of AI in the financial and stock markets sectors.
For high-risk AI applications, a combination of ex-ante assessments, based on an auditable external conformity procedure (backtesting with homogeneous requirements), as well as ex-post market surveillance would be necessary. For non-high-risk AI applications it should be allowed for companies to receive a voluntary certification.
Finally, every AI provider needs to put in place sound internal processes (i.e. modelling, training of data, handling of critical/sensitive situations, handbooks, documentation, etc.).