24/04/2026
Keeping the Friday vibe alive with our series!
We continue our Friday series where we review and announce studies that we find interesting and that we recommend investors and founders pay attention to. Today, we will be providing an overview of the report “Artificial intelligence is reshaping health systems: state of readiness across the European Union” published by WHO.
Starting with national strategies, only 3 EU Member States – Finland, Slovakia, and Sweden – have adopted a dedicated AI strategy specifically for the health sector. A further 4 countries are developing one, but the vast majority of nations have chosen instead to include health within broader cross‑sectoral AI strategies or national digital health plans. While 85% of EU Member States now have a cross‑sectoral AI strategy in place, many of these strategies remain in early stages of revision or lack a clear definition of what an AI system actually is. Oversight is equally fragmented, with most countries relying on multiple agencies or existing government bodies rather than creating new, independent entities to guide AI in health.
81% of EU Member States have consulted relevant groups on AI in health, most often through expert focus groups or informal workshops. Government actors, health care providers, and academic institutions are almost always at the table. Yet patient associations are consulted far less frequently, and the broader public is included in only four countries: Denmark, Estonia, France, and the Netherlands. Public hearings remain the least used method of engagement. Worse still, only a quarter of Member States have made the insights from their stakeholder consultations publicly available, which undermines transparency and public trust.
The health workforce is equally unprepared. Only 26% of EU Member States offer in‑service AI training for health professionals, and just 22% provide pre‑service training. Only 4 countries – Belgium, Denmark, Estonia, and Finland – offer both. When training does occur, it almost always reaches doctors, but nurses, administrators, and researchers are included far less often.
The legal and regulatory landscape is evolving but remains deeply uneven. Only 2 Member States – France and Italy – have issued health‑specific ethical guidelines for AI, and 11 countries have issued no ethical guidelines at all, nor do they plan to. Most countries are still assessing gaps in existing laws rather than creating new, dedicated AI legislation for the health sector. On liability, only 3 countries are developing specific legal standards for when AI systems cause harm; the majority are simply waiting to align with upcoming EU legislation. Alarmingly, while generative AI systems are known to have a significant environmental footprint due to high energy and water consumption, no EU Member State rated environmental impact as a major barrier to AI adoption, and 41% said it had no importance at all. Over half of Member States have regulatory agencies responsible for approving AI systems for health, but far fewer have agencies that monitor what happens after those systems are deployed in real clinical settings.
Turning to actual AI applications, the report finds that AI‑assisted diagnostics is the most common use, deployed in 74% of EU Member States, with 41% considering it fully established. This includes radiology, dermatology, and ophthalmology, where AI tools are already enhancing imaging and detection. Conversational chatbots for patient assistance come second, used by sixty‑three percent of countries, followed by automation of logistics and administrative tasks, which is active in fifty‑nine percent of Member States. Less common but growing areas include AI‑assisted prognosis prediction, surgery robotics, remote patient monitoring, and symptom checkers. Countries such as Austria, Czechia, Finland, France, Italy, Poland, and Spain are piloting or using all seven categories of AI applications.
On barriers, financial affordability is the most frequently cited obstacle, rated as being of major importance by 41percent of EU Member States. Legal uncertainty, data quality and standards, capacity gaps, lack of clear strategy, slow product approval processes, and insufficient evidence each follow at around twenty‑six to thirty‑three percent. Interestingly, legal uncertainty is a much smaller barrier in the EU than in the wider WHO European Region, likely because the EU’s AI Act and GDPR are providing a stabilising regulatory anchor. Trust and cultural impact are seen as moderately important across the board, but job displacement is considered a major concern only by Cyprus, Greece, and Italy. Most striking of all, environmental impact is not considered a major barrier by any single EU Member State, and four out of ten countries say it has no importance whatsoever – a dangerous blind spot given the well‑documented carbon and water footprint of large AI models.
When asked about policy enablers that would accelerate safe AI adoption, EU Member States spoke clearly. The single most valued measure is guidance on transparency, verifiability, and explainability of AI solutions to ensure trust in outcomes, rated as having a major positive impact by sixty‑three percent of countries. Almost equally important are accountability and liability rules for manufacturers, deployers, and users, as well as legal clarification on secondary use of health data and certification of AI systems, each rated as highly impactful by fifty‑six percent of Member States. Clarification on privacy and data protection rules for AI followed at forty‑four percent, and guidance on ethical development at forty‑one percent. Notably, no policy enabler was rated as having no positive impact, underscoring the strong consensus that deliberate, well‑designed policy is essential to making AI work safely in health care.
Our Venture Builder is looking for projects in the digital health market and is ready to help founders develop them. If you have such a project, please send your investment deck to us [email protected]
