The World Health Organization has reported that artificial intelligence is already embedded in everyday clinical work across much of its European Region, but legal and regulatory frameworks to protect patients lag far behind, according to data presented at an international conference in Lisbon.
AI in clinical practice, regulation trailing
Data from the WHO's most comprehensive survey on health-system readiness for AI — conducted across 2024 and 2025 — show that 32 of 50 responding countries (64%) now use tools that assist in diagnosis, principally to interpret imaging and identify medical findings. The survey covered responses from 50 of the 53 member states in the WHO European Region.
In addition, the WHO found that half of these countries have introduced chatbots intended to provide information, guidance and patient support. Yet regulatory preparedness is limited: only four countries (8%) reported having a dedicated national strategy for integrating AI into the health system, while a further seven countries said they were developing one. At the same low rate of 8%, only a small minority reported clear standards defining legal liability when AI systems cause an error, harm or damage to a patient.
"Questions of legal and ethical liability are no longer theoretical," the WHO data summary said, noting that AI systems are already used to mark imaging studies and triage patients.
Practical risks and policy gaps
The WHO's findings underline a growing policy gap: clinical services are adopting AI tools for image interpretation, triage and patient-facing information, but most countries have not yet settled who is responsible when an algorithm errs. That legal uncertainty extends to whether manufacturers, health facilities, individual clinicians or algorithm designers carry civil or professional liability when AI-driven decisions lead to missed or delayed diagnoses.
Policy-makers face several intertwined issues:
- Clinical safety: ensuring algorithms are validated on appropriate populations and continually audited in real-world practice.
- Legal liability: deciding responsibility for harm when human and algorithmic judgements interact.
- Transparency and accountability: setting standards for explainability, reporting of errors and post-market surveillance.
The WHO survey does not prescribe solutions, but by mapping current adoption and the paucity of national strategies it highlights where regulators must act to protect patients as AI becomes routine.
Implications for clinicians and patients
The data presented in Lisbon should prompt health authorities to consider governance frameworks that balance innovation with patient safety. Clinicians using AI-supported tools must be aware of the limits of those systems and of their professional responsibilities. Patients using chatbots or AI-driven triage should be advised to seek clinical assessment at health facilities rather than rely solely on automated advice.
Health services and regulators can draw on international experience but must tailor approaches to local legal systems and health needs. Steps that could follow from the WHO findings include strengthened requirements for clinical validation, mandatory reporting of AI-related adverse events and clarification of liability through legislation or professional guidance.
As AI expands in medicine, the WHO survey makes clear that adoption has outpaced governance in most of the European Region. For patient safety to keep pace, countries will need coherent national strategies, legal clarity and ongoing monitoring of algorithm performance in clinical settings.
Readers with concerns about symptoms or treatment should consult a doctor, clinic or their provincial health department rather than rely on online tools for diagnosis.
