International Association for Suicide Prevention

- Establish that AI governance must address mental health and all aspects of suicidality (thoughts, behaviours [including self-harm], and death) as a population-level safety issue. - Determine how AI can be made safe by design and governed to protect users, particularly those who may be thinking about or engaging in suicidal behaviour. - Establish consensus/stakeholder commitments around safety-by-design for AI platform users in crisis as an essential component. - Develop a blueprint for collaboration between other areas/entities within the Global Health Architecture. - Set up a transparent mechanism for civil society and lived-experience voices to contribute. - Establish an ethical framework for AI platforms to ensure that they are not facilitating the production of fraudulent or poor quality research on suicide prevention, which can contribute to misinformation and disinformation for the public, as well as fake evidence being used in policy and clinical practice.

• Safe, secure and trustworthy AI
• Transparency, accountability, and human oversight
• Protection and promotion of human rights
• Social, economic, ethical, cultural, linguistic and technical implications of AI

Safe, secure and trustworthy AI - chatbot and LLM safety in crisis contexts - known harms and mixed, unpredictable responses. Users of chatbots and LLMs may also be unaware of the privacy concerns of entering sensitive mental health content, which is then used to train models. Social, economic, ethical, cultural, linguistic and technical implications of AI - suicide/suicide prevention is contextual. Safety measures and AI systems trained on those contexts designed for English-speaking, high-income contexts may not be applicable elsewhere. Secondary impacts must also be considered; for example, extensive natural resource use, e.g., water for data centres, may worsen issues such as climate change, which have downstream issues for suicide prevention. Protection and promotion of human rights - the right to health, to access to care and no discrimination - especially for young people and vulnerable groups. Transparency, accountability, and human oversight - clinicians, research community, lived experience all have limited visibility on how AI responds to those expressing suicidal distress. Ethical oversight of chatbots and LLMs, where it even occurs, is opaque and processes are often proprietary. If AI is to play a more substantive role in supporting suicide prevention, greater transparency and independent oversight is essential. We know that AI is being used to mass produce unreliable and even dangerous research in mental health and suicide prevention. This research, as well as hallucinated references, is making its way into policy and practice, which risks undermining public trust in suicide prevention research.

There are possibly areas that warrant more focus; - Health in general as a cross cutting issue in terms of preventing loss of life. - Protection of young people. - AI crisis reponse.

AI is rapidly transforming the way individuals seek information, communicate, and access support. For people experiencing suicidal ideation or emotional distress, online platforms and AI-enabled tools are increasingly becoming first points of contact and opportunities for connection. Researchers document inconsistent and sometimes harmful responses from major LLMs to suicide-related prompts; and platforms deploy AI-driven detection tools with limited transparency about accuracy, bias, or support pathways. AI is increasingly being used to mass produce dangerously poor quality research and for the suicide prevention field, this can cost lives. In the same way that water companies should be held accountable for releasing raw sewage into rivers, AI companies must be held accountable for polluting the scientific suicide prevention and mental health literature and the harm this causes through misinformation, disinformation, and decaying the evidence-base. In addition, researchers are increasingly using AI in research with people experiencing suicidal thoughts and behaviours, but without clear ethical or safeguarding frameworks. AI offers significant opportunities for earlier detection of risk, improved access to support, and innovative prevention approaches, but it also presents substantial ethical, safety, and governance challenges. Ensuring that AI systems respond safely and responsibly to users experiencing distress is now a global public health priority. Furthermore, any AI-based tools for suicide risk detection, support, treatment, etc., must be open to scrutiny for high-quality, independent scientific evaluation by researchers. Proprietary algorithms and content undermine these evaluations, which are essential for determining safety and efficacy. Suicide prevention expertise is not consistently included in the development of AI safety standards, and AI expertise is not yet integrated into most national suicide prevention strategies. IASP established its Taskforce on Suicide Prevention and AI in 2026 specifically to close this gap.

Collaboration through convening Member States, UN agencies, WHO, industry, and civil society. Suicide is a highly contextual issue that requires a multisectoral response. Governance considerations cross health ministries, digital regulators, platform policy teams, and frontline services.

- WHO's work on digital health and on suicide prevention (including the LIVE LIFE implementation guidance) - IASP Taskforce on AI and Suicide Prevention IASP and partners are convening a side event at the May 2026 World Health Assembly on Online Safety, AI, Social Media, and Suicide Prevention, which is one example of the cross-sectoral collaboration.

IASP can contribute sector expertise, evidence, and pathways to lived experience Recommendations: - thematic working groups - civil society as full participants - space for sector-specific focus

#NAME?

IASPs Taskforce on Suicide Prevention and AI, spans clinical, research, policy, ethics, and lived-experience perspectives ThroughLine for crisis support: https://www.throughlinecare.com/resources/suicidal-ideation-and-self-harm-detection-in-conversational-ai Crisis Text Line & Partnership on AI work