AI Mistakes Psychologists Should Avoid in Australia (2026)

AI tools are becoming more common in Australian psychology practice, and most psychologists are finding genuine value in them. But as adoption grows, so do the mistakes — some of which carry serious professional, ethical, and legal consequences. Understanding the most common AI mistakes psychologists make is the first step to avoiding them.

This article covers the mistakes that matter most for Australian psychologists using AI in 2026, and what to do instead.

Entering Client Information into Public AI Tools

This is the most serious and most common mistake. Psychologists who use public AI tools like ChatGPT, Claude, or Google Gemini for work-related tasks sometimes include client names, dates of birth, Medicare numbers, clinical details, or other identifying information in their prompts.

This is a breach of the Privacy Act 1988 and the Australian Privacy Principles, and a potential breach of the Australian Psychological Society's Code of Ethics, which imposes strict obligations on psychologists to protect client confidentiality. It may also constitute a notifiable data breach under the Notifiable Data Breaches scheme.

The fix is straightforward: never enter any client-identifying information into a public AI tool. Use AI tools only for tasks that don't involve client data — template creation, CPD preparation, practice marketing — or use only tools that have been assessed for use with sensitive health information and that comply with Australian privacy legislation.

Signing Off on AI Documentation Without Reading It

AI tools can generate plausible-sounding clinical documentation quickly. The risk is that psychologists, under time pressure, approve AI-generated notes without reading them carefully — and errors enter the client record.

AI documentation tools can produce inaccurate clinical information, miss important details, use inappropriate clinical language, or generate content that doesn't reflect what actually occurred in the session. If you sign off on documentation you haven't read, you are professionally responsible for whatever it contains.

The rule is simple: read every piece of AI-generated documentation before it enters the client record. Correct errors. Add missing information. Make sure it accurately reflects your clinical assessment and the session content. The documentation in the client record must represent your professional psychological judgement — not just what the AI generated.

Failing to Obtain Informed Consent for AI Use

Clients have a right to know how their information is being used, including whether AI tools are being used in their care. Failing to disclose AI use to clients — or failing to consider whether disclosure is required — is an ethical oversight.

The APS Code of Ethics requires psychologists to obtain informed consent before providing psychological services and to provide clients with information that is relevant to their decision to consent. Whether AI use is material to a client's consent decision requires professional judgement — but psychologists should consider this question rather than assuming disclosure is not required.

Using AI to Substitute for Clinical Judgement

AI tools can provide clinical information, generate documentation templates, and support practice administration — but they cannot assess a client, formulate a diagnosis, or develop a treatment plan. The mistake is treating AI outputs as clinical assessments or treatment recommendations.

A psychologist who uses an AI tool to generate a treatment plan without conducting their own clinical assessment, or who relies on AI-generated clinical information without verifying it, is not meeting their professional obligations. AI is decision support — it does not replace clinical training, experience, and judgement.

Neglecting the Therapeutic Relationship

The therapeutic relationship is the foundation of effective psychological treatment. Psychologists who use AI tools in ways that undermine the therapeutic relationship — for example, by spending session time interacting with AI tools, or by using AI-generated responses in therapeutic conversations — are not meeting their professional obligations.

AI tools should be used outside of sessions, for administrative and documentation tasks. The session itself should be focused on the client and the therapeutic work.

Using Unassessed Tools with Client Data

Not all AI tools are appropriate for use with sensitive health information. Tools that have not been assessed for compliance with Australian privacy legislation, that store data on servers outside Australia, or that use client data to train their models should not be used with client information.

Before using any AI tool with client data, check its privacy policy, data processing agreements, and compliance with Australian privacy legislation. If you are unsure, seek advice from your professional association or a privacy lawyer before using the tool.

Getting the Balance Right

The psychologists who use AI most effectively are those who approach it with clear ethical boundaries — they know what AI can and can't do, they use it only for appropriate tasks, they protect client confidentiality, and they maintain their professional accountability at all times. AI is a tool that can make psychology practice more efficient and sustainable — but only when it is used thoughtfully and within the ethical framework that governs psychology practice in Australia.