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

Step 1: Pick the right tool and set expectations

How to use AI as a therapist responsibly starts with a defined self-help task and a clear understanding of the service’s limits. AI can support reflection and emotional vocabulary; a licensed clinician remains responsible for clinical care. It does not diagnose, treat, or cure mental illness, it cannot reliably 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.

For everyday concerns outside a crisis, a stress level test can help organize what you want to discuss with a professional. Grounding techniques offer a structured exercise without sharing personal information with a chatbot. These resources support reflection and skill practice; they cannot establish a diagnosis or determine immediate safety.

With that framing, choose a tool that fits your goal. Purpose-built mental-health apps may organize recognized techniques such as CBT into structured exercises. Products such as Wysa and the historically available Woebot illustrate that design approach, but clinical branding alone does not establish stronger privacy or safety. Check current availability and the evidence for the specific version. A general-purpose tool like ChatGPT is more flexible for open reflection but has weaker safety guardrails and no clinical design. Stanford HAI's June 2025 account of research presented at ACM FAccT described stigmatizing replies and unsafe responses to simulated suicidal thinking and delusions. These experiments identify failure modes; they do not establish a personal probability of harm. 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, keeps the task focused on reflection. Review generated questions and stop if they increase rumination or distress.

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.

A prompt sets a requested format; it does not guarantee that the chatbot will follow your boundaries or recognize risk. The Stanford HAI research account documents inappropriate responses despite the tools’ supportive presentation. Review what the tool actually says, and close the conversation if it supplies harmful advice. You do not need a better prompt before asking a person for help.

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 work with a therapist, an AI can fill the gaps between appointments and help you arrive better prepared. Tell your clinician which tool you use and what you ask it to do. Ask whether a paper worksheet would serve the same purpose and which topics should wait for an appointment.

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. Agree on suitable exercises with your clinician and review AI-generated summaries yourself before sharing them.

When preparing an appointment summary, ask the chatbot to distinguish events you described, feelings you named, questions you still have, and any suggestions it generated. Check each part against your own account. Keep your wording where a paraphrase changes the meaning, and label generated suggestions so your clinician can tell which ideas came from you.

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.

HHS explains that health information entered into an app for personal use is generally outside HIPAA unless the relevant service is provided by a covered entity or its business associate. Check who operates the app, what is retained, and how to request deletion. Removing names reduces disclosure but does not make a detailed life story anonymous.

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. A chatbot lacks the clinical license and professional duties of a clinician. Confident errors should be checked with someone qualified to assess your situation.

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.

A practical check before and after a chat

Before opening a chatbot, write down the task and the result you want: a clearer description of a feeling, a list of questions for an appointment, or a draft plan for an ordinary conversation. Afterward, ask whether the response helped you take a real-world step. Repeatedly asking for reassurance without feeling able to stop is a reason to pause and discuss the pattern with a professional.

For a structured alternative, use the stress level test to reflect on perceived stress and grounding techniques to practice paying attention to your surroundings. A screening result is a discussion aid, not a diagnosis or a safety assessment. The AI therapy guide covers the broader landscape, and the linked safety and prompt articles explain how these choices fit together.

Choose an ending before you begin, such as having a draft question for an appointment or a manageable action for an ordinary task. Once you have that result, save the part you want to use and close the chat. If the conversation keeps expanding, return to the original task and consider completing it on paper or with someone you trust.