AI for Therapists and Counsellors — Sensitive Practice with Sensitive Tools

Status: 🟩 COMPLETE 🟦 LIVING Section: decision-frameworks Tags: therapists, counsellors, psychologists, mental-health, decision, sensitive

The short answer

For Australian psychologists, counsellors, social workers, and mental health practitioners:

Recommended: AI for non-client work (admin, professional development, marketing)
Use carefully: AI for clinical documentation (with appropriate Australian tools and consent)
Avoid: AI for clinical decisions, AI processing identifiable client information in free tools, AI substituting for clinical judgment

Critical context: APS Code of Ethics, AHPRA Psychology Board Code of Conduct, ACA Code of Ethics, Privacy Act sensitive information, confidentiality is core to therapeutic relationship.

Mental health work has heightened ethical, legal, and practical considerations for AI. This guide explains where AI helps and where caution is essential.

The unique sensitivity

Mental health work has characteristics that make AI use particularly sensitive:

Highly sensitive information

Mental health diagnoses
Trauma histories
Family dynamics
Substance use
Relationship issues
Identity questions
Vulnerable populations

Therapeutic relationship matters

Trust foundational
Confidentiality cornerstone
AI inappropriately involved can damage trust
Clients’ awareness affects relationship

Vulnerable clients

People in crisis
Severe mental illness
Trauma survivors
Children and adolescents
People with cognitive impairment

Professional accountability

AHPRA registration (psychologists)
APS membership
ACA membership
Code of Ethics binding
Indemnity considerations

Where AI genuinely helps

Administrative work

Strongly fits:

Appointment scheduling drafts
Practice management communications
Insurance and Medicare billing language
General business correspondence
Tax and BAS preparation language

Professional development

Reading complex research
Understanding new techniques
Plain English explanations
CPD content
Reflective practice support (general, not client-specific)

Marketing and growth

Website content
Blog posts about psychology (general)
Social media for practice
Newsletter content
Referrer communications

Non-client communications

Other professionals (with confidentiality care)
Supervisor communications (anonymised)
Continuing education providers
Professional bodies

Practice operations

Staff training materials
Policies and procedures
Compliance documentation
Crisis protocols
Disaster recovery plans

Practice forms

Initial assessment templates
Consent form drafts
Intake form designs
(Specific clinical content needs professional review)

Group work preparation

General psychoeducational content
Worksheet templates (with personalisation)
Resource compilation

Where AI use is genuinely risky

Client documentation

Risks:

Patient information in AI tools
Privacy Act sensitive information
AHPRA professional standards
Insurance considerations

If using AI documentation:

Australian-built tools with Australian data residency (Heidi Health, Lyrebird Health — see those entries)
Explicit client consent
Aware of cultural considerations
Vulnerable populations particularly carefully

Caution with:

US-based tools without appropriate enterprise agreements
Free consumer AI for client notes
AI with unclear data handling

Client communications

Confidentiality care
Don’t paste identifiable info into AI
De-identify when seeking help structuring communications
Crisis communications require human judgment

Diagnostic or treatment decisions

Absolutely not:

AI making clinical decisions
AI replacing professional judgment
AI without clinical validation

AI can support:

Understanding concepts
Considering alternatives
Reflection on cases (de-identified)
(Always with professional supervision and judgment)

Crisis intervention

AI cannot:

Replace crisis assessment
Make safety decisions
Provide therapeutic relationship
Refer to services with judgment

AI must not be:

Suggested to clients in crisis as substitute
Sole crisis resource

Client-facing AI

Significant risks:

AI without clinical supervision
AI providing therapy-like responses
Inadequate crisis handling
Liability concerns

Most practitioners should NOT deploy client-facing AI without significant clinical and ethical infrastructure.

Australian professional context

Australian Psychological Society (APS)

Code of Ethics applies
Specific Ethical Guidelines
AI-related guidance evolving (2024-2026)
CPD on AI emerging

AHPRA Psychology Board

Code of Conduct for registered psychologists
General registration standards
Endorsement requirements
AI use within scope of practice

Australian Counselling Association (ACA)

Code of Ethics
Continuing professional development
AI-related guidance

Code of Ethics
Mental health social work specific
AI considerations evolving

Royal Australian and New Zealand College of Psychiatrists (RANZCP)

For psychiatrists
Medical model considerations
Specific clinical AI guidance

Indigenous mental health considerations

Cultural protocols
AIATSIS data sovereignty
Engagement with Indigenous mental health practitioners
Culturally appropriate care

Confidentiality and AI

The foundational concern:

Australian Privacy Act 1988

Health information is sensitive information
Heightened obligations
State health records legislation also applies

Professional confidentiality

Stronger than general Privacy Act standards
Clinical privilege considerations
Exception for safety risks (and other specific circumstances)
Documentation of any AI use

What NOT to put in AI tools

Client identifiable information without enterprise agreements
Session content with names
Specific case details with identifiers
Sensitive information about third parties (family, partners)

Better practices

Australian-built clinical AI with appropriate data residency
De-identification when seeking AI help
Enterprise tools with appropriate DPA
Documentation of AI use practices
Client information about AI in informed consent

If AI involvement in client work:

What AI tools are used
What information goes into AI
Where data is stored
Right to refuse AI involvement
Alternative non-AI workflow

When to disclose

AI documentation (definitely)
AI for any client-facing function (definitely)
AI in research about client populations (definitely)

Cultural considerations

Aboriginal and Torres Strait Islander clients may have specific concerns
CALD clients may have different expectations
Sensitivity to power dynamics

Vulnerable populations

Heightened consent processes
Capacity considerations
Family/carer involvement
Cultural authority

For different mental health roles

Clinical psychologists

Documentation AI (Heidi/Lyrebird with consent)
Research and professional development
Practice management
(Clinical decisions remain professional judgment)

Counsellors

Practice management AI
Marketing
Professional development
(Client work needs sensitive AI consideration)

Documentation AI within frameworks
Resource compilation
Communication drafts
Advocacy support

Psychiatrists

Clinical documentation AI
Drug information resources
Research
Letter writing
(Clinical decisions remain medical judgment)

Mental health nurses

Clinical documentation
Patient education
Communication
(Clinical decisions remain professional)

Aboriginal and Torres Strait Islander mental health workers

Cultural authority paramount
AI subordinate to traditional knowledge where relevant
AIATSIS data sovereignty
Community engagement

Researchers

AI for literature review
Analysis support
Writing assistance
(Methodology and conclusions remain researcher)

Educators in mental health

Teaching materials
Resource compilation
Case studies (de-identified)
CPD content

Practice owners

Business considerations
Staff training
Compliance
Marketing within ethical frameworks

Real workflows

Clinical documentation with Heidi/Lyrebird

Discuss AI use in informed consent
Verbal consent at session start
AI listens with consent
AI generates draft note
Practitioner reviews and edits
Note saved to practice system
Right to refuse AI: alternative workflow available

Research and professional development

AI helps understand research
Plain English explanations
Application to practice (with professional judgment)
CPD documentation
Specific to your scope of practice

Reflective practice

De-identified case discussion
AI helps consider alternatives
Supervision-style support (NOT replacing actual supervision)
Professional reflection
(Real supervision required for clinical reflection)

Practice marketing

AI helps with content drafts
Ethical promotion of services
Plain English psychoeducation (general, not personalised)
Distinct from therapeutic content

What AI doesn’t replace

❌ Therapeutic relationship — built between humans

❌ Clinical judgment — your professional training

❌ Crisis assessment — human responsibility

❌ Cultural competence — specific to community

❌ Trauma-informed care — relational

❌ Supervision — actual human supervision

❌ Personal therapy for your own wellbeing — see ai-for-mental-wellness

❌ Safety planning — collaborative human work

Specific risks

AI providing therapy-like responses to clients

Risks: incorrect responses, missed crisis, liability
Solution: avoid AI client-facing without significant clinical infrastructure

AI processing trauma content

Risks: privacy, AI training on traumatic material, content moderation issues
Solution: extreme care; enterprise tools with strong agreements

AI generating clinical documentation incorrectly

Risks: clinical errors, professional accountability
Solution: practitioner review and accountability always

AI used by clients without supervision

Risks: Character.AI, Replika, ChatGPT used inappropriately for mental health
Solution: discuss with clients; provide proper resources

AI in crisis response

Risks: missed crisis, inadequate response
Solution: AI not for crisis; human response required

Resources for Australian mental health practitioners

Professional bodies

APS (Australian Psychological Society)
AAPi (Australian Association of Psychologists Inc)
AHPRA Psychology Board
ACA (Australian Counselling Association)
PACFA (Psychotherapy and Counselling Federation of Australia)
AASW (Australian Association of Social Workers)
RANZCP (psychiatrists)
ACMHN (Australian College of Mental Health Nurses)

Regulatory

AHPRA for registered practitioners
OAIC for privacy
TGA for any medical device AI

Mental health resources for clients

Always know how to refer to:

Lifeline: 13 11 14
Beyond Blue: 1300 22 4636
Kids Helpline: 1800 55 1800 (under 25)
MensLine Australia: 1300 78 99 78
Suicide Call Back Service: 1300 659 467
13YARN: 13 92 76 (Indigenous Australians)
QLife: 1800 184 527 (LGBTIQ+)
1800RESPECT: 1800 737 732

Practice tools

Heidi Health and Lyrebird Health — Australian clinical documentation
Practice management systems with AI features
Medicare provider resources

A practical principle

The therapeutic relationship is the work. AI handles around it; never within it.

AI can transform productivity around mental health practice — admin, documentation (with consent), marketing, professional development. The clinical work itself — the relationship, the judgment, the presence, the trust — remains entirely human.

Sources

APS Code of Ethics
AHPRA Psychology Board Code of Conduct
ACA Code of Ethics
AASW Code of Ethics
RANZCP guidance
Privacy Act 1988 and state health records legislation
Lifeline, Beyond Blue, and other crisis service contact information
Personal observation of Australian mental health practice
Various international guidance on AI in mental health practice

Tech & AI, Explained

Explorer

ai-for-therapists-counsellors