In 2020, the EU Commission already emphasized in the “White Paper on Artificial Intelligence” that benefits and risks of AI are closely aligned. The use of AI may lead to violations of fundamental rights, discrimination, serious behavioural manipulation of persons and damage to health. Against this background, the proposal for a non-contractual civil AI liability aims to promote “trustworthy AI” and to make it simpler for victims of damage caused by AI to obtain compensation.
In the opinion of AK, the proposed directive requires a fundamental revision. The provisions do not protect consumers adequately at all but rather serve the interests of AI developers and AI users. Both the requirements for AI users to disclose information about the AI and the only evidentiary relief in the form of a rebuttable presumption of causality are far too high thresholds for consumers, who will not receive effective assistance as a result.