The proposed EU Directives for AI liability leave worrying gaps likely to impact medical AI

Mindy Nunez Duffourc, Sara Gerke

Research output: Contribution to journalArticlepeer-review

Abstract

Two newly proposed Directives impact liability for artificial intelligence in the EU: a Product Liability Directive (PLD) and an AI Liability Directive (AILD). While these proposed Directives provide some uniform liability rules for AI-caused harm, they fail to fully accomplish the EU’s goal of providing clarity and uniformity for liability for injuries caused by AI-driven goods and services. Instead, the Directives leave potential liability gaps for injuries caused by some black-box medical AI systems, which use opaque and complex reasoning to provide medical decisions and/or recommendations. Patients may not be able to successfully sue manufacturers or healthcare providers for some injuries caused by these black-box medical AI systems under either EU Member States’ strict or fault-based liability laws. Since the proposed Directives fail to address these potential liability gaps, manufacturers and healthcare providers may have difficulty predicting liability risks associated with creating and/or using some potentially beneficial black-box medical AI systems.

Original languageEnglish (US)
Article number77
Journalnpj Digital Medicine
Volume6
Issue number1
Early online dateApr 26 2023
DOIs
StatePublished - Dec 2023
Externally publishedYes

ASJC Scopus subject areas

  • Medicine (miscellaneous)
  • Health Informatics
  • Computer Science Applications
  • Health Information Management

Fingerprint

Dive into the research topics of 'The proposed EU Directives for AI liability leave worrying gaps likely to impact medical AI'. Together they form a unique fingerprint.

Cite this