AI-systems pose new legal challenges to our approach to liability. The boundaries are still not clear, and the existing framework remains a valid reference, but we must be able to accompany the rapid development of this emerging technologies. We probably need a differentiated approach for the time being, starting with the definition of the various levels of risk that AI applications might imply. The report overall contains a balanced position, including also considerations about a mandatory insurance for high-risk AI-systems. I voted in favour.