How to Use AI as a Therapist: A Practical, Responsible Guide

Step 1: Pick the right tool and set expectations

Before anything else, be clear about what AI is and is not. AI is a self-help and emotional-support tool, not a licensed therapist. It does not diagnose, treat, or cure mental illness, it cannot assess risk, and it is not a crisis service. If you are in crisis or thinking about suicide, call or text 988 in the US to reach the Suicide and Crisis Lifeline, available 24 hours a day.

With that framing, choose a tool that fits your goal. Purpose-built mental-health apps like Wysa or Woebot are built around recognized techniques such as CBT and tend to handle risk and privacy more carefully than a general chatbot. A general-purpose tool like ChatGPT is more flexible for open reflection but has weaker safety guardrails and no clinical design. A 2025 Stanford study presented at ACM FAccT found chatbots responded unsafely to mental-health scenarios about 20 percent of the time, compared with roughly 7 percent for human therapists. Match the tool to the job, weigh the AI therapy pros and cons before you commit to one tool, and read its privacy policy before you share anything.

Set expectations up front. Decide what you want from a session, whether that is venting, untangling a decision, or practicing a skill. Treat the AI as a thinking partner that reflects your words back, not as someone who understands you or is responsible for your well-being.

Step 2: Write good prompts and ask focused questions

The quality of what you get back depends heavily on how you set up the conversation. Give the AI a clear role and clear limits in your first message, then keep your questions specific.

For guided journaling: "Act as a reflective journaling guide. Ask me one open question at a time about how my day went. Do not give advice unless I ask. Keep your responses short." This kind of guided journaling, sometimes called AI journal therapy, is one of the safest ways to start.

For naming a feeling: "I feel off but cannot name it. Ask me a few questions to help me figure out what emotion I am experiencing and what might have triggered it. Do not diagnose me."

For CBT-style reframing: "Here is a thought that is bothering me: 'I always mess everything up.' Help me examine it like a CBT worksheet would. Name the likely thinking distortion, ask me for evidence for and against it, and help me write a more balanced thought."

For breaking down overwhelm: "I am overwhelmed by everything on my plate. Help me list it out, sort by what is actually urgent, and pick one small next step. Just one."

Ask one thing at a time rather than dumping everything at once, and push back when the answer feels too agreeable. Chatbots tend to tell you what feels good to hear, so ask it to challenge your thinking, not just validate it. For more ready-made AI therapist prompts, see our examples guide.

Step 3: Use it between real sessions and to prepare for them

AI works best as a supplement to care, not a replacement for it. If you already see a therapist, an AI can fill the gaps between appointments and help you arrive better prepared. This is already common: in a 2026 American Psychological Association survey, more than a third of psychologists reported patients using AI as an extra mental-health professional.

Use it to organize what you want to raise next session. One good prompt turns a messy week into a focused agenda: "Help me organize what I want to talk about with my therapist this week. Ask me what has been on my mind, then summarize it into three clear points."

Between sessions you can practice the skills a clinician taught you, rehearse a hard conversation in a low-stakes way, or do a quick check-in when something flares. Bring anything meaningful back to your therapist, who can weigh it in context the AI does not have. The AI handles the homework. The human handles the care.

Step 4: Track patterns over time

One genuine strength of using AI consistently is that it can help you notice trends you would otherwise miss. A short daily or weekly check-in builds a record of your mood, triggers, and what helps.

Ask it to help you reflect, then keep your own notes outside the chat. For example: "Each evening, ask me to rate my mood from one to ten and name one thing that affected it, then keep your follow-up short." Over a few weeks you can ask it to help you look back and spot themes, such as poor sleep tracking with low mood, or a recurring worry showing up before work.

Remember the AI usually does not carry memory across conversations reliably, so do not depend on it to remember your history or a safety plan. Hold the longer view yourself, in a journal or notes app, and share patterns with a professional when they matter.

Step 5: Protect your privacy

Everything you type into an AI is data, and these conversations touch some of the most sensitive information about you. Treat them as not private by default.

Avoid sharing full names, addresses, employer details, or anything that identifies you or other people. Use general terms instead of specifics where you can. Before you commit to a tool, read how it handles and stores data, whether it trains on your inputs, and whether you can delete your history. Purpose-built mental-health apps are often clearer about this than general chatbots, though practices vary widely.

If you would not want a piece of information stored on a server or seen by someone else, do not put it in the chat. Privacy is a feature you should choose for, not an afterthought.

Guardrails: what AI is not for

Knowing where to stop is the most important part of using AI responsibly. AI is not for crisis. It cannot keep you safe, assess how serious your situation is, or respond reliably when you are at risk. If you are thinking about harming yourself or others, contact a licensed professional or, in the US, call or text 988, available 24 hours a day.

AI is not for serious or worsening mental illness. Conditions like severe depression, bipolar disorder, psychosis, trauma, abuse, or an eating disorder need a trained clinician, not a chatbot. AI is also not a clinician: it is not licensed, owes you no duty of care, and answers to no one when it sounds confident but gets things wrong.

Use AI as a first step or a supplement for everyday reflection and skill-building. These steps track the emerging best practices for AI chatbots in therapy. The moment things get heavy, persistent, or risky, move to a human. Still asking yourself, is AI therapy safe? We answer that in a dedicated guide. And if you would rather talk to a person, browse licensed therapists in our directory.