Balancing AI’s Potential and Peril in Health
WHO Europe has issued a warning about the rapid adoption of artificial intelligence across health systems, noting clear benefits alongside serious risks. AI tools can speed diagnosis, streamline workflows, and support clinical decision making, but those gains come with hazards that affect patient safety and privacy.
The WHO report highlights three core concerns:
- Data quality and bias: AI models trained on incomplete or unrepresentative data may produce biased, inaccurate, or unsafe recommendations. Biased outputs can amplify disparities in care and harm patients who are already underserved.
- Privacy vulnerabilities: Patient records and imaging data are highly sensitive. WHO notes that fewer than one in five countries have national guidance on health data use for AI, leaving personal information exposed to misuse and reidentification risks.
- Overreliance and infrastructure gaps: Clinicians may over-trust automated outputs, especially where clinical governance, training, and digital infrastructure are weak. Inadequate oversight and poor integration of AI into workflows increase the chance of error.
Urgent Need for Regulation and Trust
Governance gaps are wide. The report states that only about 28 percent of countries have ethical rules for companies developing AI. That shortfall creates accountability blind spots for developers, purchasers, and regulators.
Trust remains a key barrier. High-profile examples of biased algorithms in other sectors have made clinicians and patients cautious about entrusting critical decisions to machines. Rebuilding trust requires transparent validation, independent testing, and mechanisms for redress when systems fail.
WHO calls for coordinated action: clear legal frameworks, independent safety evaluations, robust privacy safeguards, mandatory human oversight, and investment in workforce training. Policymakers, health system leaders, and vendors must cooperate to make sure AI delivers benefits equitably and does not erode patient safety or dignity.
This is a time for vigilance rather than unchecked rollouts. Responsible policy and practical standards will determine whether AI is a tool for better care or a source of new harms.
