WHO publishes ethics and governance guidance for large multi-modal models in healthcare

The World Health Organization released guidance on 18 January 2024 covering over 40 recommendations for the development, regulation, and deployment of large multi-modal models (LMMs) in healthcare contexts. The guidance identifies five healthcare use cases (diagnosis, patient self-triage, administrative, education, research) and documented risks including hallucination, data bias, accessibility barriers, and automation bias. The document proposes new regulatory frameworks, including third-party auditing, post-deployment impact assessments, and stakeholder-inclusive design processes. For financial institutions, this signals an emerging policy consensus that AI healthcare applications will face mandatory impact assessments and regulatory approval workflows, raising compliance and deployment costs for vendors serving health systems.