Home » Insights » AI and Digital Ethics in Psychology Practice: What the HPCSA Now Requires

AI and Digital Ethics in Psychology Practice: What the HPCSA Now Requires

ai-ethics-psychology-practice-hpcsa

Artificial intelligence is no longer a distant concept in healthcare — it is already shaping how psychologists document sessions, communicate with clients, and make clinical decisions. Understanding AI ethics in psychology practice is now a professional requirement, not an optional interest. In November 2025, the Health Professions Council of South Africa updated Booklet 20 to provide explicit guidance on the ethical use of AI for all registered health practitioners. If you are an HPCSA-registered psychologist, this booklet applies directly to you. You can explore our HPCSA-accredited CPD packages to meet your ethics CEU requirements for the current cycle.

What Is HPCSA Booklet 20 and Why It Matters for Psychologists

On 20 November 2025, the HPCSA released a revised version of Booklet 20 — Ethical Guidelines on the Use of Artificial Intelligence (Health Professions Council of South Africa, 2025). This update marks the first time the HPCSA has set out a comprehensive framework for how AI tools may be used in clinical and administrative practice in South Africa.

The booklet applies to all HPCSA-registered practitioners, including psychologists across all registration categories — counselling, clinical, industrial, educational, and research. Furthermore, it covers a wide range of AI applications: from clinical decision-support tools and diagnostic aids, to AI-assisted documentation and patient engagement platforms.

Critically, the guidelines do not prohibit AI use. Instead, they establish the conditions under which AI may be used ethically and responsibly. Practitioners who ignore these requirements risk professional accountability, regardless of whether the AI tool caused direct harm.

The Three Pillars of Ethical AI Use Under the HPCSA

Booklet 20 organises ethical AI use around three pillars. Each pillar carries distinct obligations for psychologists in practice (Health Professions Council of South Africa, 2025).

Pillar 1: Ethics

All AI tools used in your practice must support — not undermine — core ethical principles. This includes respect for patient autonomy, privacy, and confidentiality. Specifically, the HPCSA requires that the standard of care remains consistent whether or not AI is involved. Patients who decline the use of AI tools may not be disadvantaged or refused access to services as a result of that refusal.

Additionally, only AI that has been validated for reliability and cultural appropriateness may be used. This is a particularly significant requirement in the South African context, where many AI tools have been trained primarily on Western, English-language datasets that may not be representative of local populations.

Pillar 2: Legal

Any AI platform used in your practice must comply with South African legislation. The Protection of Personal Information Act (POPIA) is the primary legal framework governing how client data is collected, stored, and processed. Consequently, before adopting any AI tool that handles client information, you must verify that the platform is POPIA-compliant and that adequate data processing agreements are in place (Health Professions Council of South Africa, 2025).

The Promotion of Access to Information Act also applies. Moreover, international platforms — even widely used ones — do not automatically meet South African legal standards. Practitioners remain accountable for verifying compliance before use.

Pillar 3: Technical

AI tools must have defined criteria, features, and capabilities. Systems must be capable of meeting the intended clinical or administrative objectives. Furthermore, they must meet cybersecurity, data quality management, and interoperability standards. This means that adopting an AI tool because it is popular or convenient is not sufficient. You are required to evaluate its technical integrity before integrating it into your practice.

What These Guidelines Mean in Day-to-Day Practice

For most psychologists, the practical implications of Booklet 20 centre on three common scenarios: AI-assisted session notes, telehealth platforms with built-in AI features, and psychological assessment tools that use algorithmic scoring or interpretation.

In each of these scenarios, your core obligation is the same: you remain professionally accountable for every clinical decision, regardless of what an AI tool has suggested, generated, or recommended. As the APA’s 2025 guidance states, AI should augment — not replace — human decision-making, and psychologists must retain the authority to approve or reject AI-generated recommendations based on their own professional judgement (American Psychological Association, 2025).

Therefore, if you use an AI scribe to transcribe or summarise sessions, you are responsible for reviewing and correcting that output before it enters the clinical record. If an AI tool generates a risk-assessment summary, you are responsible for evaluating it critically before acting on it.

Ready to earn your ethics CEUs and stay current with HPCSA requirements? Browse the 2026 CPD packages at PsyCampus — or find out if you qualify for a recent graduate discount of up to 40%.

AI and Informed Consent in Psychology

One of the clearest obligations introduced by both the HPCSA and international professional bodies is the requirement to obtain informed consent before using AI tools in a client’s care. This applies whether the AI tool is used for documentation, clinical decision support, or any other purpose that involves the client’s personal information (American Psychological Association, 2025).

Informed consent in this context means more than a signature. It requires that you explain, in plain and culturally appropriate language, what the AI tool does, what data it accesses, how that data is stored, and what the client’s right to decline looks like in practice. This conversation must happen before the tool is used — not after the fact.

Additionally, where a client declines the use of an AI tool, the HPCSA is explicit: that client may not be disadvantaged or refused access to services. You must have a non-AI pathway available for all core clinical functions.

Managing the Risks: Bias, Over-Reliance, and Data Privacy

The HPCSA acknowledges that AI tools carry genuine risks, and Booklet 20 specifically flags three areas of concern: algorithmic bias, over-reliance on automated systems, and data privacy vulnerabilities (Health Professions Council of South Africa, 2025).

Algorithmic bias is a particular concern in South Africa. Many AI tools have been developed and validated on datasets from high-income, predominantly Western contexts. Consequently, their reliability for South African clients — across language, culture, socioeconomic background, and diagnostic presentation — cannot be assumed. Practitioners are expected to critically evaluate whether a tool is appropriate for the specific population they serve.

Over-reliance is equally problematic. A peer-reviewed review published in December 2025 noted that mental health professionals are increasingly uncertain about how to navigate ethical decision-making in the AI era, particularly when AI-generated outputs appear authoritative (Grzybowski et al., 2025). The safeguard is straightforward: treat all AI outputs as preliminary inputs, not final conclusions.

On data privacy, any tool that processes client information must be evaluated for POPIA compliance before adoption. This includes cloud-based platforms, AI scribes, and any software that transmits data outside of South Africa’s borders.

AI Ethics in Psychology Practice: Staying Compliant Going Forward

Booklet 20 is now a live professional standard. All HPCSA-registered practitioners are expected to familiarise themselves with its requirements, and that expectation applies immediately — not at the next renewal cycle (Health Professions Council of South Africa, 2025).

Practically, this means conducting an audit of any AI tools currently in your practice, verifying their POPIA compliance, updating your informed consent documentation, and putting a process in place to critically review AI-generated outputs before they influence clinical decisions or enter the clinical record.

Furthermore, governance does not stop with you individually. If you supervise other practitioners or manage a practice, you are responsible for ensuring that staff are trained and that ethical protocols around AI are actively implemented and reviewed.

Staying current with the HPCSA’s evolving standards is exactly what CPD is designed to support. If you are looking to meet your ethics CEU requirements while building practical knowledge for your work, explore the PsyCampus 2026 CPD packages — accredited, relevant, and designed to fit your registration category.

References

Scroll to Top