Implementation guide for ethics in health artificial intelligence systems
Guide d’implémentation de l’éthique dans les systèmes d’intelligence artificielle en santé
This implementation guide provides 43 concrete criteria organized by the chronological development phases of health AI systems to help providers integrate ethical values such as beneficence, non-maleficence, autonomy, justice, and eco-responsibility. Developed by the Ethics Cell of the Digital Health Delegation, it combines stakeholder collaboration, comparative analysis with existing ethical labels, and alignment with the European AI Act. It aims to support trustworthy, human-centric AI innovation in healthcare.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
This guide is intended for providers, developers, and deployers of health artificial intelligence systems seeking practical methodologies to ensure ethical compliance.
What the document contains
The document addresses the integration of ethical principles into the lifecycle of health AI systems, detailing criteria for data collection, algorithmic design, user interfaces, deployment, and continuous evaluation, alongside self-assessment tools for providers.
It does not provide legally binding regulations as it serves as a guidance tool rather than an enforceable law, though certain criteria may feed into future sector-specific frameworks.
- The guide establishes 43 operational criteria structured across the framing, development, deployment, and continuous improvement phases of health AI systems.
- It emphasizes the preservation of human autonomy, clinical critical thinking, and robust human oversight mechanisms over automated AI outputs.
- The methodology includes comparative benchmarking against major European and national AI labels and alignment with the European AI Act.
- It provides concrete self-assessment elements and examples for each criterion to assist providers in demonstrating ethical compliance.
- Special attention is given to generative AI risks, such as hallucination, plagiarism, and the leakage of personal data.
· Éditorial de l’expert · Éditorial des pilotes du GT3 · Membres du groupe de travail GT3 · 1. Introduction · 2. Méthode de travail adoptée par le GT3 · 3. Version finale du guide d’implémentation de l’éthique dans les SIA en santé · 4. Conclusion · 5. Annexe 1 : Comparaison de la V0.1 du guide avec les labels de l’éthique de l’IA · 6. Annexe 2 : Règlements et normes associés aux critères du guide d’implémentation et propositions d’éléments d’auto-évaluation
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Human Dignity
4/5Respect for the inherent worth of persons
C’est un lieu de confiance, de fragilité, de responsabilité partagée. Dans ce lieu, toute innovation doit d’abord se plier à une exigence : celle de servir l’humain, non de s’y substituer.
Éditorial des pilotes du GT3
screening score · 2026-10-25