Proposals for a new legal framework on ai: Possible discrepancies between liability, compliance and duties of company directors

Abstract

Two recent proposals for an EU Regulation on AI, both of which adopt a risk-based approach, should be aligned with one another. However, the no-fault liability regime concerning high-risk AI systems appears to be disconnected from the duties of compliance inspired by the Ethical Guidelines for Trustworthy AI. Furthermore, the duties of company directors, and in particular the duty to act on an informed basis and the duty to overview when using non-high-risk AI systems for important decision-making, could be a concrete obstacle to the complete success of these proposals. In this regard, a human-centric approach should help to solve these issues.