You've probably heard about AI chatbots like ChatGPT, or noticed mental health apps popping up in your app store. These tools are becoming more common, and they can do some genuinely useful things, but what does that actually mean for your wellbeing, and how should you think about using them?
01 · The landscapeWhat this is about
AI tools can chat with you around the clock, help you think through what you're feeling, suggest coping strategies, and offer a kind of emotional support, often for free or at a very low cost. That's a real shift in what's available between appointments. This handout is here to help you use them in a way that supports your wellbeing rather than complicates it.
02 · Two different thingsApps vs. AI chat engines
Before we go further, it helps to know that not all "AI" is the same. Two very different kinds of tools tend to get lumped together, and it matters which one you're using.
Mental health apps
Purpose-built for wellbeing (for example, mood trackers, CBT apps, guided meditation, sleep support). They're usually trained on a specific, curated set of materials such as clinical frameworks, evidence-based exercises, and content reviewed by clinicians. Most also have guardrails built in.
General AI chat engines
Tools like ChatGPT, Gemini, Claude, or Copilot. These are trained on huge amounts of text from across the internet and are designed to talk about almost anything. That makes them flexible, but it also makes them much harder to keep safe for a specific purpose like mental health.
The safety rails built into an AI
Guardrails are the rules and safety checks a developer builds into an AI to shape how it responds: what it can say, what it must refuse, when it should hand you off to a human, and how it deals with sensitive topics like self-harm, medication, or crisis.
- In a mental health app, guardrails are usually tighter and more specific: content is limited to what the app was designed for, crisis triggers link to local helplines, and prompts stay within an evidence-based framework.
- In a general AI chat engine, guardrails are broader and harder to tune. Because the tool has to handle any topic, safety rules can be inconsistent, easier to talk around, and vary from one engine to the next.
- No guardrails are perfect. Both kinds of tools can still miss things, give unhelpful advice, or respond in ways their developers didn't intend.
Something else worth knowing: different chat engines can give you quite different answers to the same question. Each one is trained on different material, tuned in different ways, and carries different built-in biases (cultural, clinical, commercial). One engine might steer you towards a coping skill, another towards seeing a professional, and another towards something less helpful entirely. If a response feels off, it's worth checking a second source or, better, checking in with a person.
03 · The upsideWays AI can actually help
Let's start with the good stuff. AI tools have some real advantages. They're available whenever you need them (3 AM or during your lunch break), and they don't judge. If you want information about anxiety, depression, or stress, you can find it instantly without waiting for an appointment. Cost is another big one, especially if you're managing on a tight budget or can't easily afford appointments. And honestly, sometimes you just need a quick thought at an odd hour, and AI can be there for that moment.
Available around the clock
Late-night worries, early-morning racing thoughts, the middle of a difficult afternoon. AI is there without you having to wait.
Low-cost information
Quick, plain-language explanations of what anxiety, depression, or stress actually are, and simple strategies to try.
A non-judgemental space
Somewhere to put words on what you're feeling before you're ready to say it to another person.
A thinking partner
A place to hold a passing thought, question, or reflection until you can talk it over with someone who knows you.
04 · The honest bitWhat AI cannot do
Here's where we need to be clear. AI is not a mental health support professional, and it can't do the things one does. What good mental health support offers is human connection, trust built over time, and someone who knows you, your history, and your specific needs. A person who knows you can pick up on what you haven't said, understand the nuances of your situation, and adjust their approach based on who you are. AI cannot do this. It doesn't have clinical judgement, it doesn't know you, and it can't adapt to the complexity of your life the way a person can.
A few safety details worth remembering
If an AI detects that you're in crisis, it may stop responding or direct you to emergency services. This is actually a safety feature, but it means you shouldn't rely on AI when you're in real distress.
- AI can be confidently wrong. Chat engines sometimes invent facts, misquote research, or give advice that sounds convincing but isn't right for your situation. If it matters, check a second source or a person.
- AI is not a diagnostic tool. Don't use it to decide what condition you have, whether to start, stop, or change medication, or whether something is or isn't serious. Those calls belong with a qualified professional.
- Overuse is a real risk. Chatbots are designed to keep you engaged. If you find you're turning to AI more than to people, or using it late into the night to soothe difficult feelings, that's worth noticing and talking about.
Think of AI more like a well-read friend who's good at organising information, not a stand-in for the people in your life who actually know you.
05 · Fit for purposeWhat AI is actually good for
So where does AI fit in your mental health picture? It can be genuinely useful for learning coping strategies, like breathing exercises or ways to challenge unhelpful thoughts. It's helpful for understanding the basics of how your mind works, or what anxiety and depression actually are. And it can be a good space for practising self-reflection and exploring your thoughts without pressure.
06 · Cultural contextSomething most people don't mention
AI tools are trained mostly on Western, English-language information. That means the advice, frameworks, and ways of thinking they offer are built on Western values and assumptions. If you come from a different cultural background, or if your spirituality, community structures, or ways of managing wellbeing don't fit the Western mental health model, an AI might not get it. What it suggests might not fit your circumstances, your values, or what actually works for your life and culture.
This is a real limitation, and it's another reason it helps to talk with someone who knows your context, whether that's a professional, an Elder, a community leader, or a person who shares your background.
07 · When it's not mental health at allCircumstances vs. conditions
AI can't always tell the difference between a mental health condition and a reasonable reaction to extraordinarily difficult circumstances. If you're struggling because you've experienced a disaster, lost your home, or are living in poverty, that's not a mental illness that needs a coping strategy. That's a human reaction to a situation that needs practical help. You might need housing, food, income support, legal aid, or community resources, not a chatbot telling you how to manage your stress better.
AI won't reliably recognise this difference, and it might suggest you need to "fix" yourself when what you actually need is for your circumstances to change.
Reach for a person, not just a chatbot
If you're using AI and something doesn't sit well, or you suspect what you're dealing with is more about your circumstances than your mental health, reach out to a person or service that can help with what's actually happening. A counsellor, a community worker, a local service, or someone you trust can help you work out whether you need mental health support, practical assistance, or both.
08 · Data & privacyWhat happens to what you share
It's completely reasonable to wonder where your words go once you've sent them. The honest answer: most AI services keep a copy of what you say, and some use it to help train future versions of themselves. Free tools tend to be the least private. Paid or clinical-grade tools sometimes offer more protection, but not always, and it's not always obvious which is which.
A simple rule of thumb
AI chats may store your conversations, whether you're typing or speaking. Avoid sharing sensitive details unless you're sure the app is private.
A good rule: talk to an AI the way you might talk in a busy café. You can share how you're feeling, work through a problem, or think out loud, but leave out the identifying stuff: full names, addresses, phone numbers, dates of birth, exactly where you work or live, and the same for anyone you're talking about. Keep it general and it stays useful without becoming a record you might not want sitting on someone else's server. "I've been anxious about something at work" is plenty for the AI to help you with. "I've been anxious about [colleague's name] since the meeting on Tuesday" is more than it needs.
The same thinking applies to voice: talking is quicker and feels more natural, but the words still get sent to a server, and often they're transcribed and stored just like text. If it's something you'd hesitate to put in an email, it's probably not something to say out loud to an AI either.
If what you want to talk about touches on trauma, abuse, medication, legal matters, or anything you'd want kept private, a person is a safer place for it than a chatbot, ideally your mental health support professional or GP. If none of them are within reach right now, a crisis or helpline counsellor (see section 10) is bound by confidentiality in a way an AI simply isn't. If you do end up using AI in the meantime because it's what you have, keep the details general and the identifying stuff out of it.
09 · A conversation worth havingIf you have someone you can tell
If you're using AI for mental health support and you have a mental health support professional, tell them about it. Yes, really. Then talk about it together:
- Is what the AI is saying actually helpful, or making things more confusing?
- Are there suggestions that don't feel right for you?
- Could it be interfering with your care in any way?
- How could you and your professional use it as a shared tool rather than a separate one?
The best approach, if you have one, is to work through those questions together, so you're both clear on whether, and how, AI is genuinely supporting your care.
10 · When you need real peopleCrisis and acute distress
If you're in a crisis, in acute distress, or having thoughts of harming yourself, this is the time to reach for actual human support. AI isn't equipped for this, and you deserve real, immediate human connection when things are that hard.
Reach a real person now
Call a crisis line, contact your mental health support professional or GP, go to your nearest emergency department, or call someone you trust. All lines below are free and available 24/7.
13YARN supports Aboriginal & Torres Strait Islander callers. Kids Helpline is for ages 5–25. 1800RESPECT is the national sexual assault, family & domestic violence counselling service.
A note for young people and parents: if you're under 18, or supporting someone who is, please reach for a trusted adult, school counsellor, GP, or Kids Helpline rather than relying on an AI chatbot. Australia's eSafety Commissioner has cautioned that AI companion apps can feel supportive but are not designed for young people's safety.
11 · If supports aren't availableWhat if there's no one nearby?
For many people, especially in isolated or rural areas, finding professional support can be incredibly difficult. There might be no local professionals, services can be hours away, or there's simply not enough funding for mental health services in your area. If that's your situation, here's what's real: AI can help you hold the gap while you keep looking for human support. It isn't a substitute, but it can be a useful bridge.
If you're in isolation with minimal local supports, you can use AI as a thinking partner. It can help you work through problems, learn about what you're experiencing, and practise strategies when there's no one else to talk to right now. It might help you feel less alone, even though you know it's not a real relationship. That matters, but it's a bridge, not a destination.
Keep checking in with yourself
Is this actually helping, or am I just feeling heard by a tool that isn't really understanding me?
- If what you're dealing with is practical (food, housing, legal help, income support), AI can help you find information about services, but it can't solve the underlying problem.
- Telehealth can bring a counsellor, GP, or psychologist to your phone or laptop, which opens up services well beyond your local area.
- See if there are online communities of people in similar situations.
- If possible, try to connect with at least one real person: a phone call with a distant counsellor, a video session, a community health worker on their next visit, a GP, or a trusted friend or family member. Real human connection, even occasional and imperfect, matters in ways AI cannot replicate.
You might also reach out to services that support rural and remote communities specifically. Many organisations understand isolation and have adapted their services to reach people where they are, even if it's not traditional face-to-face support.
Useful tool, not a substitute
AI can be a helpful addition to your toolkit, a supportive presence when you need quick information or a thought partner. It works best when it sits alongside the people in your life, not in place of them.
AI is at its best when it's one thread in a bigger web of support: people who know you, professionals if you can reach them, and tools like this used with a bit of care. Your wellbeing matters, and how you weave those threads together is what tends to matter most.