WHO Calls for Stronger Oversight as AI Expands Into Health Policy

The paper was developed jointly by WHO’s Department of Data, Digital Health, Analytics and AI and the Department of Science for Health.

The World Health Organization (WHO) has released a new discussion paper highlighting the growing role of artificial intelligence (AI) in health policy-making and the need for stronger safeguards to ensure decisions remain grounded in reliable evidence.

While much of the global debate around AI has centred on its use in clinical care and medical technologies, WHO says another important area deserves attention: the way AI is increasingly shaping health policies. The organisation argues that AI is beginning to influence how health problems are identified, how policy options are developed, and how the success of those policies is measured.

According to WHO, these changes create opportunities for faster analysis and improved decision-making, though they also raise concerns about transparency, fairness and accountability.

Risks and opportunities emerge across every stage of policy development

The paper was developed jointly by WHO's Department of Data, Digital Health, Analytics and AI and the Department of Science for Health. It is designed to support policy-makers, regulators, health managers and AI developers as they navigate the growing use of AI in government decision-making.

WHO organises its assessment around the different stages of the policy cycle, including understanding public health challenges, designing solutions, implementing policies and evaluating results. AI can help process large amounts of information, identify trends and support evidence reviews. At the same time, the organisation warns that risks can appear at every stage.

Biased datasets may lead to inaccurate problem assessments. Systems optimised for measurable outcomes could overlook broader social needs. Unequal access to digital technologies may limit the effectiveness of implementation efforts. Monitoring tools may also introduce hidden biases that gradually influence policy outcomes.

The paper highlights a particular concern known as "epistemic injustice," where AI systems give greater weight to data that can be easily measured while overlooking lived experiences, Indigenous knowledge, local expertise and community perspectives.

WHO stresses that AI should support, not replace, human judgment

WHO says countries do not need to start from scratch when developing safeguards for AI. The paper points to existing frameworks and principles that already emphasise transparency, public participation, human rights protections and risk-based oversight.

Practical recommendations include conducting algorithmic impact assessments before AI systems are adopted, reviewing the readiness of technologies, maintaining human verification of evidence gathered by AI tools and establishing multidisciplinary oversight groups that include technical, policy and ethics experts.

The organisation emphasises that human judgement must remain central to policy-making. AI can help analyse larger datasets, support ongoing evidence reviews and improve scenario modelling, but people should continue to define policy questions, evaluate evidence quality, interpret findings and make final decisions.

WHO views the discussion paper as the beginning of a broader conversation with governments, researchers, developers and civil society groups. As AI technologies continue to evolve, the organisation says ongoing dialogue will be essential to ensure innovation strengthens evidence-informed policy-making rather than undermining public trust.