WHO Turns Its AI Scrutiny on Policy-Making Itself: New Paper Sets Guardrails for Evidence Work

The World Health Organization has published a discussion paper — "Artificial intelligence and evidence-informed policy: emerging challenges and opportunities" — examining AI's role across the three stages of health policy-making: problem definition, solution design, and implementation and monitoring.

The paper's core observation, articulated by WHO's Dr. Alain Labrique, is that the global AI-in-health debate has fixated on clinical care while neglecting a quieter question: how AI is starting to shape the evidence base that policy itself is built on — the literature reviews, syntheses, and monitoring data that determine what governments do.

The proposed guardrails

WHO's recommendations include algorithmic impact assessments before deployment, technology readiness reviews, "living evidence workflows" that pair automated retrieval with human verification, and multidisciplinary oversight panels. The through-line: humans retain responsibility for framing questions and weighing ethics, with AI in an accelerant role built on existing tools such as GRADE and the OECD AI Principles.

The paper extends WHO's prior guidance — its 2021 AI ethics framework and 2024 guidance on large multi-modal models — into the machinery of policy-making, and arrived ahead of WHO/Europe's "Shaping AI in Health" conference in mid-July.

For health ministries and, increasingly, for US agencies building AI into regulatory workflows, the message is the same one WHO has delivered on clinical AI: the technology can be an analytical support tool, but it cannot be the decision-maker.

Reporting: World Health Organization, June 2, 2026.