# WHO Forms AI Ethics Consortium in Healthcare with Focus on Mental Health
Delft University of Technology (TU Delft) in the Netherlands has become the first WHO Collaborating Centre for AI governance in healthcare, including ethical aspects. This marks the beginning of the Consortium of Centres on AI and Health—a network of institutes covering all six WHO regions. The goal is to develop unified standards for AI use in medicine. The inaugural meeting of candidates took place March 17–19 in Delft, where participants agreed on priorities and collaboration mechanisms.
The initiative stemmed from a seminar on January 29 ahead of the India AI Impact Summit 2026 under WHO auspices. Over 30 experts in AI, mental health, ethics, and policy highlighted the impact of generative AI on public mental well-being. Particular attention was paid to chatbots used for psychological support without proper design and testing.
Alain Labrique, director of WHO's Department of Digital Health and AI, emphasized: AI interacts with users in moments of emotional vulnerability, so systems require development and regulation with safety in mind.
Recommendations for Assessing AI Risks to Mental Health
Experts identified three priorities for integrating mental health into AI regulation:
- Assess the impact of generative AI on mental health in any decision-making, not just in specialized therapy apps.
- Monitor long-term effects, including emotional dependence on AI systems.
- Develop mental health support tools in collaboration with psychiatrists, people with lived experience of mental disorders, including youth, taking into account cultural and linguistic differences.
These measures are based on research. A study in the Journal of Medical Internet Research found that youth aged 18–25 at risk of psychosis more frequently use AI chatbots for emotional support and anthropomorphize them, attributing therapist roles. TU Delft researchers propose standardizing crisis handoff protocols: chatbots should redirect users in acute states to specialists.
Regional Coverage and Global Standards
The consortium will ensure participation from institutes across all WHO regions, adapting standards to local needs. This is not a ban on AI in healthcare, but rather building an evidence base for ethical governance. For AI developers, it means shifting from national norms to global protocols, especially in mental health.
Tech companies will face requirements for testing psychological effects. For example, chatbots for everyday conversation need mechanisms to detect crises and escalate to professionals. Implementing such protocols minimizes risks from uncontrolled use of generative models.
Key Points
- TU Delft is the first WHO consortium center for AI in healthcare with an emphasis on ethics.
- Generative AI affects mental health through off-label use of chatbots for therapy.
- Recommendations: comprehensive risk assessment, dependence monitoring, collaborative development with experts.
- Crisis handoff protocols are needed to redirect users to specialists.
- The consortium will cover all six WHO regions for global standards.
— Editorial Team
No comments yet.