A trade show turned showcase of transformation
For its sixtieth edition, held from 19 to 21 May 2026 at Paris Expo Porte de Versailles, SantExpo confirmed its status as the annual gathering of healthcare and medico-social stakeholders. Organized by the Fédération hospitalière de France (FHF) and run by GL events, the trade show brought together, according to its official report, more than 37 180 participants (nearly 6 % more than in 2025), nearly 700 exhibitors and around 750 speakers during 304 presentations. It was inaugurated by the Minister of Health, Stéphanie Rist, and by the President of the Senate, Gérard Larcher, under the theme “Excellence in health: a commitment for all generations.”
Among the topics on the agenda (prevention, attractiveness of professions, mental health, ecological transition, autonomy), artificial intelligence occupied a central place, with an agora entirely dedicated to “AI in health.” Above all, the tone has changed. For years, the discourse on medical AI was about foresight. In 2026, the question asked is more demanding: what does AI actually do for patients, caregivers and institutions, at what cost and with what risks?
A national strategy taking shape
At the Ministry of Health stand, the Délégation au numérique en santé (DNS) presented the “Artificial Intelligence and Health Data” strategy2025-2027, organized around five axes: governance involving field experts, mapping of uses, evaluation of systems integrating AI, clarification of the regulatory framework and adaptation of training. A guide from the HAS and the CNIL on the proper use of AI systems in care settings was then under public consultation.
The first figures illustrate the appetite of institutions. The call for expressions of interest “AI & Planning, AI & Emergency Departments,” launched in 2025 by the DGOS and the ANAP, received 339 applications, of which 204 were selected. According to the DNS, half of the projects were already being deployed or in use three to four months later. The feedback from the Valenciennes hospital center on predicting emergency department flows summed up the spirit of the moment: the initial gains are mainly organizational, and deployment is first and foremost a matter of governance and skills.
Other funding supports this movement. On 9 April 2026, the Minister of Health announced funding of 14,9 million euros from France 2030 for Alliance Santé IA, presented as the first sovereign AI integration program in a university hospital, led by Montpellier University Hospital and ADLIN Science. At the European level, the Commission opened on 21 April 2026 seven calls for projects under the Digital Europe Programme, for a total of 63,2 million euros. This envelope does not only concern health: it also covers digital skills and online security. The health component includes 9 million euros for AI-assisted imaging screening and around 24 million for digital health services related to the European Health Data Space.
Health data, treasure and Achilles' heel
At the heart of all these issues lies health data, both the raw material of medical AI and a major point of vulnerability. To train models useful for diagnosis or research, considerable volumes of structured, quality clinical data are needed. But this data is among the most sensitive there is.
This is the whole challenge of the European Health Data Space (EHDS), created by Regulation (EU) 2025/327. Entered into force on 26 March 2025, it applies from 26 March 2027, then in stages: cross-border exchange of the first categories of data, such as patient summaries and electronic prescriptions, is planned from March 2029, and most of the rules on the reuse of data for research will also apply from 2029. The hosting issue remains sensitive: data stored on infrastructures subject to extraterritorial laws poses a problem of trust and strategic autonomy.
Cybersecurity is the other side of this concern. Healthcare institutions are regular targets of cyberattacks, and each new AI system expands the attack surface. A medical AI, however effective, becomes a danger if the data it processes can be stolen or its operation can be hijacked.
Clinical AI vs. administrative AI
A debate runs through the sector: should AI first serve care or management? Clinical AI promises to improve diagnosis and accelerate research, but its deployment faces considerable scientific and regulatory validation requirements. Administrative AI, less spectacular, can relieve caregivers more quickly: coding of procedures, schedules, reports. In a system where caregiver time is the scarcest resource, its impact can be more immediate.
General practitioner Vincent Bacquey, digital general officer of the Military Health Service since 1st March 2026, expressed this measured hope in an interview published by the Ministry of the Armed Forces in May: “The medtech sector is booming with numerous innovations. There is great hope in the use of new technologies to further increase the quality of care.”
The validation puzzle
Medical devices incorporating AI fall under both the European Medical Device Regulation and the AI Act, which classifies them as high-risk systems. However, the timeline has shifted: the AI omnibus, which entered into force on 27 July 2026, postponed to 2 August 2028 the application of high-risk rules for AI integrated into regulated products such as medical devices. This delay gives industry time, but does not exempt them from articulating two demanding regimes. The balance between safety and innovation takes on particular urgency here, because error is measured in human lives.
Ethics and the risk of a fig leaf
How far can we delegate to an algorithm decisions that affect patients' health? How can we prevent data biases from becoming discrimination? These questions went beyond the trade show. On 21 May, in Nice, the Club Santé de Nice-Matin and the EDHEC Chair in Management in Innovative Health organized a round table, moderated by journalist Nancy Cattan, on a related theme: how to move “from promises to uses” and transform innovation into real impact in health.
One risk remains: presenting AI as a response to the hospital crisis, marked by caregiver shortages and bed pressures. Automating coding or optimizing schedules can provide marginal relief, but will not replace human and financial investment in the healthcare system. Used as a fig leaf, AI would become a means of doing ever more with ever less.
From experimentation to proof
SantExpo 2026 gave the feeling of a sector changing phase. After years of prototypes, the task now is to demonstrate, with supporting data, that AI brings measurable value. This proof phase requires rigorous validation protocols, independent evaluation and transparency about the limitations of tools. It also requires resolving the issues of sovereignty, cybersecurity and ethics that underpin trust. Next appointment: SantExpo 2027, from 25 to 27 May in Paris.
Sources: Official SantExpo report 2026 (Ehpadia, 11 June 2026) ; Ministry of Health, AI agora SantExpo 2026 ; Hospitalia, Alliance Santé IA ; European Commission, 21 April 2026 ; European Commission, EHDS regulation ; Ministry of the Armed Forces, interview with MG Vincent Bacquey, 11 May 2026 ; European Commission, AI omnibus ; Invest in Côte d’Azur, EDHEC round table of 21 May 2026.
