Your clients are already using artificial intelligence for mental health support — whether you know about it or not. A 2025 peer-reviewed study published in the APA journal Practice Innovations found that among people using AI for mental health support, 87% had prior experience with human therapy — and nearly 75% rated their AI experience as equal to or better than human therapy (Rousmaniere et al., 2025).
Note: This study was conducted among US adults. No equivalent South African study currently exists, though the global trend is consistent with emerging usage patterns across the African continent (FINN Partners, 2025).
This guide gives you a verified breakdown of the current AI mental health landscape in South Africa, what the evidence says about these tools, what your professional obligations are when clients use them, and what South Africa’s new regulatory developments mean for your practice. For CPD content that addresses digital practice ethics — including the 5 compulsory ethics CEUs required annually — explore the PsyCampus CPD packages for your registration category.
The AI Mental Health Landscape in South Africa
South Africa faces a significant mental health care gap. The World Health Organization estimates fewer than five mental health professionals are available per 100,000 people globally — a gap that is substantially more acute in South Africa’s public sector context, where mental health services remain under-resourced relative to population need (Botes, 2025). Specifically, digital mental health tools offer three things the traditional system often cannot: immediate access, affordability, and anonymity.
Globally, mental health applications now number in the tens of thousands. In the South African context, tools commonly used include:
- Mindfulness and meditation apps — Headspace, Calm, and Insight Timer are widely used in South Africa for stress reduction and emotional regulation (Botes, 2025)
- AI chatbot therapy tools — Wysa and Woebot use AI to deliver cognitive behavioural therapy (CBT) prompts, mood tracking, and emotional support in conversational formats
- Vimbo — a South Africa-based app offering self-guided digital therapies specifically for depression and anxiety in the African context (RGA, 2025)
- Teletherapy platforms — platforms connecting users to licensed therapists via chat, video, or voice, which gained significant traction during and after the COVID-19 pandemic
- General-purpose AI assistants — a growing number of people use tools such as ChatGPT for emotional processing, journalling, and informal mental health support, despite these tools not being designed or validated for clinical use
Furthermore, a pilot using AI-powered chat in local South African dialects showed encouraging user engagement — reflecting both the demand for and the potential of culturally responsive digital mental health support in the South African context (FINN Partners, 2025).
South Africa’s Landmark AI Mental Health Regulatory Framework
In November 2025, at the G20 Social Summit in Johannesburg, South Africa announced a landmark national initiative — the Comprehensive AI Regulation and Evaluation for Mental Health (CARE MH) programme. Developed by the South African Health Products Regulatory Authority (SAHPRA) and PATH, with funding from Wellcome, CARE MH is the world’s first regulatory framework specifically designed for AI tools in mental health (PATH, 2025).
Specifically, CARE MH will establish a science-based and ethically robust framework describing how AI mental health tools must be evaluated for safety, inclusivity, and effectiveness before receiving market authorisation in South Africa. Consequently, this development signals that South Africa is moving towards formal regulation of AI mental health tools — and that the current landscape, where many apps operate without rigorous clinical validation, is set to change.
For psychology practitioners, this has two immediate implications: the AI tools your clients are currently using may not yet have been evaluated under any formal framework, and your professional obligation to assess these tools’ appropriateness remains fully active in the interim.
What the Evidence Actually Says About AI Mental Health Tools
Understanding the evidence base is essential for having informed conversations with clients about these tools. A systematic review of 160 studies from 2020–2024, published in PMC in 2025, provides the most current and comprehensive picture (PMC, 2025):
| Finding | What it means | Clinical implication |
|---|---|---|
| LLM-based chatbots now represent 45% of new studies (2024) | The field has shifted rapidly toward generative AI chatbots | Tools are evolving faster than the evidence base |
| Only 16% of LLM-based chatbot studies included clinical efficacy testing | Most AI therapy tools have not been tested for actual symptom reduction | Do not recommend unvalidated tools as clinical interventions |
| Rule-based chatbots (Woebot, Wysa) show strongest clinical evidence | Structured CBT-based chatbots have the most rigorous evidence base | These tools are most defensible as adjuncts for mild-moderate presentations |
| Many tools marketed as “AI-powered” use simple rule-based scripts | Marketing claims often exceed actual AI capability | Evaluate tools independently rather than relying on marketing |
| Digital forms of talk therapy reduce symptoms of anxiety and depression for mild to moderate cases | There is genuine evidence of benefit — but specifically for lower-severity presentations | AI tools are not appropriate for clinical-level or crisis presentations without human oversight |
The PoPIA Problem: What Clients Are Risking When They Use Mental Health Apps
Data privacy is one of the most serious and least visible risks associated with AI mental health tools. A 2025 peer-reviewed study in PMC specifically addressing South Africa’s regulatory challenges found that popular AI-driven therapy apps including BetterHelp and Talkspace had shared user conversations and mental health assessments with third-party advertisers without clear consent (Botes, 2025). Mental health data is among the most sensitive personal information that exists — and many apps process it without adequate PoPIA-compliant safeguards.
Specifically, South Africa’s current legal framework — including PoPIA, the National Health Act, and the Consumer Protection Act — does not yet provide sufficient clarity on how mental health data collected through AI-driven applications must be handled (Botes, 2025). Consequently, users of these tools face real and current data privacy risks, and practitioners have a professional responsibility to inform clients of these risks when they disclose using such tools.
Your Professional Obligations When Clients Use AI Mental Health Tools
Ask about AI use as part of your intake process
Given the prevalence of AI mental health tool use among people who are also in therapy, incorporating a question about digital mental health tool use into your intake assessment is now professionally sound practice. Specifically, understanding which tools a client is using, for what purpose, and how frequently gives you clinically relevant information about their help-seeking behaviour, their digital literacy, and any potential risks from unvalidated tools.
Be able to discuss these tools knowledgeably
Practitioners who dismiss AI tools without engaging with the evidence risk alienating clients and creating distance in the therapeutic relationship (Schneider, 2026). Furthermore, clients who use AI tools between sessions may benefit from guidance on which tools have stronger evidence bases, how to use them appropriately for self-support, and what the limitations are. This requires practitioners to engage with the landscape rather than dismiss it.
Clearly distinguish your role from AI tools
When clients use AI tools alongside therapy, establishing clear boundaries around what each serves is clinically important. AI tools are appropriate for self-support, psychoeducation, mood tracking, and mild-to-moderate self-help between sessions — they are not appropriate for crisis support, clinical assessment, diagnosis, or therapeutic intervention with complex presentations. Moreover, a client in crisis reaching out to an AI chatbot instead of a human practitioner or crisis line is a genuine clinical risk that requires proactive conversation.
Apply HPCSA guidelines when recommending digital tools
HPCSA Booklet 20 requires that only AI tools validated for reliability and cultural appropriateness in the South African context be used in healthcare practice (Health Professions Council of South Africa, 2025). Consequently, if you recommend a digital mental health tool to a client as part of their care — even informally — you take on a degree of professional responsibility for that recommendation. Recommending general-purpose AI assistants or unvalidated apps in a clinical context constitutes an ethical risk under the updated HPCSA guidelines.
AI as a Complement, Not a Replacement
The most defensible and evidence-based position for South African psychology practitioners in 2026 is clear: AI mental health tools are potentially useful complements to human-provided care for mild-to-moderate presentations — they are not substitutes for professional psychological practice, and they carry real risks that clients need to understand.
South Africa’s development of the CARE MH regulatory framework signals that the formal landscape is evolving rapidly. As a result, staying current with how these tools are being regulated, validated, and used in the South African context is increasingly part of professional competence — and directly relevant to your ethics CPD obligations.
For a full overview of how AI ethics obligations fit within your HPCSA ethical framework, read our guide: AI and Digital Ethics in Psychology Practice: What the HPCSA Guidelines Say. To meet your 2026 CPD requirements in a format that is flexible, accredited, and category-specific, browse our CPD packages — or check whether you qualify for a recent graduate discount of up to 40%.
References
- Botes, M. (2025). Regulatory challenges of digital health: The case of mental health applications and personal data in South Africa. Frontiers in Digital Health. Retrieved from https://pmc.ncbi.nlm.nih.gov
- FINN Partners. (2025). AI and mental health in Africa: A quiet digital shift is making therapy more accessible. Retrieved from https://www.finnpartners.com
- Health Professions Council of South Africa. (2025). Booklet 20: Ethical guidelines on the use of artificial intelligence. Guidelines for Good Practice in the Health Professions. Retrieved from https://www.hpcsa.co.za/ethics
- PATH. (2025). South Africa, PATH, and Wellcome launch world’s first AI framework for mental health at G20 Social Summit. Retrieved from https://www.path.org
- PMC. (2025). Charting the evolution of artificial intelligence mental health chatbots from rule-based systems to large language models: A systematic review. Retrieved from https://pmc.ncbi.nlm.nih.gov
- RGA. (2025). AI chatbots break down barriers to much-needed mental health treatments. Retrieved from https://www.rgare.com
- Schneider, J. G. (2026). Why are clients turning to AI for mental health? Psychology Today South Africa. Retrieved from https://www.psychologytoday.com/za
- PsyCampus. (2026a). AI and digital ethics in psychology practice: What the HPCSA guidelines say. Retrieved from https://www.psycampus.co.za
- PsyCampus. (2026b). HPCSA accredited CPD courses for psychologists. Retrieved from https://www.psycampus.co.za