Two-Thirds of European Countries Use AI in Medical Diagnosis

Artificial intelligence has quietly become standard practice in European healthcare systems, with nearly two-thirds of countries now deploying AI tools for medical diagnosis. Yet the regulatory frameworks meant to protect patients when these systems fail remain largely absent, according to new data from the World Health Organization.

The findings, presented at an international conference in Lisbon, stem from the most comprehensive survey the WHO has conducted on health system readiness for AI. The organization collected responses from 50 of 53 member states in its European region during 2024 and 2025.

Adoption outpaces oversight

Thirty-two countries—64% of respondents—reported using AI tools that assist in diagnosis, primarily for interpreting medical images and identifying clinical findings. Half have introduced chatbots designed to provide patient information, guidance, and support.

The contrast with regulatory preparedness is stark. Only four countries, representing 8% of respondents, have established a dedicated national strategy for integrating AI into healthcare. The same percentage have defined clear legal liability standards that specify who bears responsibility when an AI system causes patient harm.

This gap is no longer theoretical. An AI system might flag an X-ray as normal despite a visible tumor, underestimate a patient's deteriorating condition, or assign low priority to a case requiring urgent care. While physicians typically make final decisions, the question of accountability grows more complex as these systems become deeply embedded in clinical workflows.

Why it matters

The liability vacuum creates immediate risk for patients and legal uncertainty for healthcare providers. With 86% of countries identifying legal ambiguity as the primary barrier to safe AI adoption, health systems face a difficult choice: deploy tools that may improve diagnosis and efficiency, or wait for regulatory frameworks that may take years to develop. Meanwhile, patients are already receiving care influenced by algorithms with no clear recourse when errors occur.

Training and bias concerns

Staff preparation lags behind deployment. Only one in five countries provides AI training to healthcare professionals before they enter the workforce. One in four offers structured training for current workers. Less than half have assessed whether existing legislation adequately covers AI use, and nearly 40% lack ethical guidelines for the technology in healthcare settings.

Bias represents another significant risk. AI systems trained on databases that underrepresent women, minorities, older adults, or patients with rare diseases may perform less accurately for these populations. An algorithm successful in one hospital may behave differently under different workload conditions or in another country.

Benefits driving adoption

Despite these concerns, 98% of countries cited improved patient care as their primary motivation for adopting AI. Ninety-two percent sought to reduce burden on medical staff, and 90% aimed to improve efficiency. In Portugal, image analysis systems help identify chest diseases and bone fractures faster, reducing wait times in primary care and emergency departments.

The WHO emphasized that systems must not become "black boxes." Physicians need to understand accuracy rates, testing populations, failure modes, and appropriate responses when AI recommendations conflict with clinical judgment. Patients require transparency about AI involvement in their care and clear pathways for addressing harm.

Privacy concerns add complexity, as these systems require extensive medical data including diagnoses, medications, genetic tests, and mental health records. The WHO called for clear rules governing data collection, security, secondary use, and patient rights.

Dr. Hans Kluge, WHO Regional Director for Europe, warned that the gap between implementation and oversight could become irreversible, particularly as patients increasingly consult chatbots about symptoms before speaking with doctors. The organization urged countries to build national strategies, establish liability standards, invest in training, and ensure real-world testing beyond laboratory conditions.

These details were first reported by The Jerusalem Post, based on data presented at the WHO conference in Lisbon attended by ministers and senior representatives from 37 countries across all six WHO regions.