Skip to main content
Please wait...

 Database index

Soft law

WHO LMMs Guidance

Full name of the document
World Health Organization Ethics and governance of artificial intelligence for health: Guidance on large multi-modal models
Body / organism enacting the document
World Health Organization (WHO)
Nature of the body / organism
International body
Type of document
Project area
AI and health
AI and data protection
Law area
Civili liability
Privacy / data protection
Consumer protection
Fundamental rights protection
Product liability
Product safety
Medical devices

Summary of the document

The recommendations could be divided into two categories depending on the stakeholders to whom they are addressed.

The key recommendations for the governments are:
1.    Investment in public infrastructure and datasets for AI development, demanding adherence to ethical principles.
2.    Enforcement of laws and regulations ensuring that LMMs in healthcare meet ethical and human rights standards.
3.    Designation of regulatory bodies to assess and approve LMMs for healthcare, incorporating post-release auditing and impact assessments by independent entities. These evaluations should be public and consider various user demographics.

The key recommendations for the developers are:
1.    Involvement of a broad range of stakeholders, including potential users and healthcare professionals, early in the AI development process to address ethical concerns and enhance application relevance.
2.    Focus on designing LMMs for specific, beneficial tasks in healthcare, ensuring accuracy, reliability, and the capacity to anticipate secondary outcomes.

Type of addressees
Governments and developers
Territorial scope
Situations involved

WHO has issued new guidance focused on the ethics and governance of LLMs in healthcare. LMMs, which can process and generate diverse outputs from various data types like text, videos, and images, have seen rapid adoption and offer potential to enhance healthcare delivery. The guidance encompasses over 40 recommendations targeting governments, technology companies, and healthcare providers to ensure LMMs are used effectively and ethically to promote public health and address health inequities.

AI system(s) involved
  • Specific intended purpose AI
  • Generative AI
  • Foundation model
  • Deep learning
Fundamental rights involved
  • Environmental rights
  • Right to data protection
  • Right to health
  • Right to privacy
  • Right to private and family life
  • Right to informed consent
  • Right to non-discrimination
Principles expressly addressed
  • Accountability
  • Equality
  • Explainability
  • Non-discrimination
  • Transparency
Connection with hard law

There’s not an explicit referral to hard law instruments, however, the Guidance should be implemented and used as a model for any national laws specifically regulating healthcare, access to healthcare data, data protection (especially biometric data), medical research, cybersecurity (particularly of medical devices), consumer protection and liability. 

Possible legal force and impact on national/supranational legal system

Countries looking to develop or update their legal frameworks to include AI in healthcare may use the WHO guidance as a foundational document. It can help in drafting laws that reflect global standards and best practices in the ethics and governance of LMMs.

WHO's guidance might influence future international agreements or declarations (particularly on regional level) related to digital health and AI, setting a precedent for the ethical use of LMMs in healthcare on a global scale.

Case author
Ivo Emanuilov
LIBRe Foundation