Healthcare AI 2025

FRANCE Trends and Developments Contributed by: Liliana Eskenazi, Julie Ernewein and Pauline Lecrenais, Fréget Glaser et Associés

Ensuring compliance with growing EU and national regulations From a legal standpoint, the deployment of AI in healthcare is now governed by a set of overlapping European and national regulations including: • the AI Act (Reg. 2024/1689); • Medical Devices Regulations (MDR) and In Vitro Diagnostics Regulations (IVDR); • the European Health Data Space (EHDS) Regula - tion; • the Product Liability Directive (PLD); • the General Data Protection Regulation (GDPR) and the French Data Protection Act; and • Article L. 4001-3 of the French Public Health Code. Under the AI Act, AI-based medical devices are clas - sified as high-risk systems, requiring dual conformity under both the AI Act and the MDR/IVDR. Guidance issued in June 2025 by the Medical Device Coordina - tion Group (MDCG) establishes coordination rules to avoid duplicative procedures. While the Act’s provisions concerning high-risk AI sys - tems will come into effect in August 2026, the Europe - an Commission has not yet published its guidelines on classifying high-risk AI systems and related require - ments and obligations. On this subject, two recent developments are worth mentioning, as follows. • In June 2025, the European Commission launched a public consultation on high-risk AI systems under the AI Act seeking stakeholder input to refine defi - nitions, classifications, and regulatory obligations. According to the European Commission, “this feedback will be taken into account in the upcom - ing Commission guidelines on classifying high-risk AI systems, and related requirements and obliga - tions. It will also collect input on responsibilities along the AI value chain”. • Leading EU companies issued an open letter requesting a two-year delay for certain AI Act obli - gations (notably for general-purpose and high-risk AI), citing legal uncertainty and competitiveness concerns. The EU Commission immediately stated that there will be no postponement and no pause in the roll-out of its AI regulations.

More broadly, AI is now increasingly recognised as a strategic lever for transforming the French healthcare system in response to structural challenges such as workforce shortages, ageing populations and rising costs. Nonetheless, the implementation of AI in healthcare remains subject to considerable challenges. Key con - cerns include data quality, algorithmic bias and lack of transparency, as well as ethical risks such as the potential dehumanisation of care and unequal access to innovation. Legal and regulatory hurdles, including liability issues and the absence of dedicated reim - bursement models, also hinder broader deployment. From an economic perspective, high development and deployment costs remain a major barrier. Furthermore, user acceptance is far from assured. According to the 2024 barometer published by PulseLife and Interaction Healthcare, only 58.7% of healthcare professionals trust AI for diagnostic tasks. Among their concerns, algorithmic bias (59%), lack of transparency about data sources (50%), and potential degradation of the professional–patient relationship (49%) are most frequently cited. On the patient side, an OpinionWay survey for the Healthcare Data Insti - tute shows that only 44% of patients believe doctors can safely use AI in medical care. Lastly, the regulatory environment surrounding AI in healthcare is rapidly evolving. In addition to new EU regulations – some of which are already entering into force – numerous guidelines have been published or are forthcoming from key institutions such as HAS, the Digital Health Delegation, and the CNIL. Stakeholders must therefore remain attentive to legal developments at both national and European levels. Key Precautions in Developing and Using Health AI in France In light of the evolving regulatory environment, one of the primary challenges facing developers and users of AI systems in the healthcare sector is to understand how the new European instruments interact with one other and with existing national legal frameworks.

CHAMBERS.COM