Many people now use AI tools like ChatGPT between therapy sessions. They may use them to journal, organise thoughts, seek reassurance, practise difficult conversations, or make sense of emotions. Feeling nervous about telling your therapist does not mean you have done something wrong. Often, the most useful question in therapy is not simply whether you use AI, but what role it is playing emotionally and psychologically in your life.

AI may offer reassurance, language, or reflection, but it cannot replace genuine human connection. Psychological wellbeing is shaped not only by information and advice, but by meaningful relationships, social bonding, belonging, and the experience of being known by another person.

A growing number of people are quietly using AI for emotional support.

For some, it starts almost accidentally. They type out their thoughts after an argument. They ask ChatGPT to help them organise what they want to say in therapy. They use it to journal late at night when their mind will not slow down. Others ask questions they feel too embarrassed to ask another person.

Sometimes the AI use feels practical. Sometimes it feels comforting. Sometimes it becomes more emotionally important than the person expected.

Then another thought appears:

Should I tell my therapist about this?

That question can feel surprisingly loaded.

Some people worry their therapist will think they are replacing therapy with technology. Others fear sounding emotionally dependent, “weird”, needy, or too attached to an AI tool. Some already suspect the AI use may be becoming unhelpful, but feel ashamed admitting how much they rely on it during moments of distress.

At Stronger Minds, therapy is understood as more than advice-giving. It is a place where coping strategies, emotional patterns and relationship difficulties can be explored carefully, including the ways people now use digital tools between sessions.

The important question is rarely just:

“Is AI good or bad?”

It is more often:

“What is this doing for you psychologically?”

That question can open up meaningful therapeutic work.

Myth: “My Therapist Will Judge Me for Using AI”

This is probably one of the most common fears.

Many clients imagine that bringing up AI in therapy will offend the therapist or damage the therapeutic relationship. They worry it will sound as though they do not trust the therapy, or that they are trying to find a shortcut around difficult emotional work.

But therapy is rarely helped by secrecy.

If somebody says, “I’ve been using ChatGPT when I panic at night,” that may tell us something important about overwhelm, emotional loneliness, difficulty self-soothing, or fear of burdening other people.

If somebody says, “I ask AI whether I’m overreacting after arguments,” that may point toward reassurance-seeking, uncertainty, interpersonal anxiety, or difficulty trusting their own judgement.

And if somebody says, “It feels emotionally safer than talking to people,” that may become an important doorway into shame, attachment, rejection, emotional exhaustion, or previous relational experiences.

A psychologically informed therapist is usually less interested in policing coping strategies and more interested in understanding them. If you are unsure what kind of professional support would best fit your needs, it may help to understand the difference between psychologists, counsellors, coaches and psychiatrists.

That does not mean all AI use is healthy. It also does not mean all AI use is harmful. Therapy is usually more nuanced than that.

Did you know?
A strong therapeutic alliance is consistently associated with better psychotherapy outcomes across different therapy approaches, although it is not the only factor that contributes to change (Flückiger et al., 2018).

Myth: “Using AI Means Therapy Isn’t Working”

People sometimes assume that if they are using AI between sessions, it must mean therapy is failing.

Occasionally, it may be a sign that something needs to be reviewed. Perhaps the sessions are not addressing the right issues. Perhaps the person does not yet feel safe enough to be fully open in therapy. Perhaps they need more structure between sessions. If this feels familiar, it may be worth reading more about what it can mean when therapy is not helping.

But often it is more complicated than that.

Therapy takes place within the limits of ordinary human life. Sessions end. Therapists are not available twenty-four hours a day. Emotional crises do not politely wait for appointment times.

Some people turn to AI during moments of distress because they feel emotionally flooded and need somewhere to put their thoughts. Others use it because they struggle to organise emotions internally before speaking. Some are trying to reduce the intensity of their thoughts before bringing them into therapy.

In these situations, AI may sometimes function a little like a reflective notebook, a sounding board, or an emotional organising tool.

However, there is another side to this.

Certain kinds of AI use may unintentionally reinforce the very psychological patterns somebody is struggling with. This is especially relevant when someone repeatedly uses AI to seek certainty, check whether they are “right”, confirm whether they are a good person, or reduce anxiety after every uncomfortable thought or interaction.

Repeatedly asking, “Do you think I handled this badly?”, “Am I a narcissist?”, “Does this sound like OCD?”, or “Do you think they secretly hate me?” may provide temporary relief while strengthening the cycle underneath.

That distinction matters clinically.

A behaviour can feel soothing in the short term while still maintaining anxiety over time. For people who recognise this pattern in anxiety, it may be useful to learn more about anxiety symptoms and therapy, because reassurance-seeking can become one way anxiety keeps itself going.

The aim is not to shame people for using AI. It is to understand what emotional process is happening around the AI use itself.

Myth: “AI Is Either Helpful or Dangerous”

Public conversations about AI and mental health often become strangely extreme.

Some people speak as though AI is revolutionising therapy. Others describe it as inherently dangerous or emotionally toxic.

The reality is more nuanced.

AI may sometimes help people organise thoughts, journal, practise language, or feel temporarily reassured. For some, it may offer a useful bridge between therapy sessions. It may also be especially appealing for people receiving online assessment and therapy, because digital tools may already feel like a normal part of how they reflect, communicate or prepare.

But reassurance is not the same as relationship.

This distinction matters clinically.

Human beings do not develop psychological resilience only through information, advice, or repeated reassurance. We also adapt through relationships. We learn emotional regulation through being responded to by other people. We develop trust through repeated experiences of being heard, misunderstood, repaired with, challenged, accepted, and held in mind by another person.

That is one of the reasons therapy can be powerful. At its best, therapy is not only a place where someone receives techniques, explanations, or coping strategies. It is also a relational space where emotional experience is shared with another human being.

AI may imitate validation, warmth, curiosity, or reassurance. It may even feel easier than talking to a person because it is immediate, predictable, and non-judgemental.

But it cannot create genuine social bonding. It cannot offer the felt experience of belonging to another human relationship. It cannot participate in the mutual, embodied, emotionally meaningful process through which loneliness is often softened.

Irrespective of how much reassurance someone receives from an AI model, it cannot replace the therapeutic effect of real human connection.

This does not mean AI has no value. It means its value needs to be understood within limits.

AI may support reflection.

But reflection is not the same as relationship.

And reassurance is not the same as belonging.

Did you know?
Loneliness and social disconnection are associated with poorer mental and physical health outcomes. Social connection is increasingly understood as a major determinant of wellbeing, not just a pleasant extra (Holt-Lunstad, 2024).

What Therapists Are Often Actually Concerned About

Many people assume therapists are mainly worried about being replaced.

In practice, therapists are often more concerned about whether AI use is helping someone move towards life, relationships, and emotional flexibility or whether it is becoming a way of retreating from them.

Some people describe AI as feeling emotionally safer than human beings because it is instantly available, consistently validating, predictable, and non-confrontational. Psychologically, that can make sense. If someone has repeatedly experienced criticism, rejection, emotional neglect, inconsistency, bullying, trauma, or relational instability, a predictable AI response may feel deeply relieving.

But therapy may eventually need to ask a deeper question:

“Is this helping you become more connected to yourself and others, or is it helping you avoid the risks of human connection?”

That question is not an attack on the person. Often, it is a compassionate attempt to understand an emotional survival strategy.

The reassurance someone receives from AI may reduce distress briefly. But it cannot provide the same developmental and therapeutic benefits as meaningful human contact. It cannot offer reciprocal care, shared emotional presence, rupture and repair, facial expression, tone of voice, embodied warmth, or the gradual experience of being known by another person.

This is especially important when loneliness, shame, rejection sensitivity, trauma, low self-worth, or attachment difficulties are part of the picture. Some people who struggle with emotional intensity, relational insecurity, or self-worth may also recognise themes discussed in therapy for personality difficulties, although AI use itself should never be used to make assumptions about diagnosis.

A person may feel soothed by AI and still remain profoundly lonely.

That is why a therapist may want to explore not only whether AI feels helpful, but whether it is helping the person move towards deeper human connection — or away from it.

Why It Can Feel So Difficult to Mention AI in Therapy

The discomfort many people feel is often not really about technology.

It is about vulnerability.

Saying, “I’ve been talking to AI about my emotions,” can carry a surprising amount of shame.

For some people, it touches fears of sounding emotionally needy, being seen as dependent, not coping properly, being judged, appearing socially isolated, or being misunderstood.

Sometimes people even apologise while describing it.

But those emotional reactions are often clinically meaningful in themselves.

A client may say, “I didn’t want to burden anyone.” Another may say, “It feels easier than talking to real people.” Someone else may say, “I feel embarrassed needing reassurance.” Another may admit, “I only use it when I’m spiralling.”

These are not meaningless details. They may point toward deeper themes involving attachment, emotional safety, shame, self-worth, loneliness, or fear of being too much for other people.

This overlaps with a broader pattern many people recognise: wanting help, but fearing that asking for help will burden others. If that resonates, this Stronger Minds article on feeling like a burden and how to talk anyway may be a useful companion piece.

In some cases, bringing AI use into therapy may deepen the therapeutic work rather than interfere with it.

Healthy vs Unhelpful AI Use

There is not yet a universally agreed clinical framework for understanding AI use in therapy.

But one useful question is:

“What happens psychologically after I use AI?”

For some people, AI may help create enough emotional distance to reflect more clearly. It may support journalling, emotional awareness, or preparation for therapy conversations. That kind of reflective use can sometimes sit alongside therapy in a helpful way, rather like writing thoughts down before a session.

For others, the interaction gradually becomes more compulsive. The person notices they increasingly rely on AI for reassurance, certainty, emotional regulation, or validation. Instead of helping reflection, it starts feeding over-analysis and emotional dependency.

Usually the distinction is not completely black and white.

The important issue is less whether somebody uses AI and more whether the pattern increases reflection or avoidance, emotional awareness or emotional escape, flexibility or dependency, groundedness or compulsive reassurance.

Another important question is:

“Is this helping me move towards meaningful human contact, or away from it?”

This is not about forcing people to talk to others before they feel ready. For some, AI may be a stepping stone. It may help them rehearse language, clarify what they feel, or prepare to speak more openly in therapy.

But if AI becomes a substitute for human contact, the longer-term costs may be significant. Psychological adaptability in the real world depends on the ability to form, maintain, tolerate, and repair meaningful human relationships. These relationships are not always easy. They involve uncertainty, disappointment, negotiation, vulnerability, and sometimes conflict.

But they also create belonging.

AI can provide a response.

It cannot provide belonging.

This is where therapy differs from a purely informational exchange. To understand that distinction more deeply, it may help to read about what clinical psychology is and why psychological therapy involves more than advice-giving.

Did you know?
In psychological therapy, behaviours are often explored according to the function they serve emotionally rather than whether they are simply “good” or “bad”.

How to Bring It Up in Therapy

Many people imagine they need to justify or carefully explain their AI use before mentioning it.

Usually they do not.

Often, a simple and honest opening is enough:

“I’ve actually been using ChatGPT a bit between sessions.”
“I feel awkward mentioning this, but I’ve been talking to AI when I’m anxious.”
“I think I might be using AI for reassurance sometimes.”
“I wanted to explore whether this has been helping or making things worse.”
“I’ve noticed AI feels easier than talking to people, and I’m not sure what to make of that.”

You do not need to arrive with a polished conclusion.

Part of therapy may involve working that out together.

A good therapist is unlikely to need you to present the issue perfectly. More often, they will be interested in understanding what happens before, during, and after the AI use. They may ask what you tend to seek from it, how you feel when it responds, whether it makes you feel more able to face life, or whether it leaves you needing more reassurance.

They may also help you think about whether AI is supporting the therapy or quietly replacing some of the human contact that your mind and body may need.

This is also why the fit between client and therapist matters. If you feel anxious about how a therapist will respond to something vulnerable, you may find it helpful to read about choosing between a male or female therapist and what affects emotional safety in therapy.

That conversation can be uncomfortable.

But it can also be useful.

When AI Use Is Connected to Relationships, Arguments or Loneliness

Many people do not only use AI for general emotional support. They use it after something has happened with another person.

They may paste in a message from a partner, friend, colleague, or family member and ask, “Was I wrong?” They may describe an argument and ask AI to decide who was being unreasonable. They may ask it to draft a reply when they feel too angry, hurt, anxious, or ashamed to think clearly.

There can be value in slowing down before responding. But relationships are rarely solved by outsourcing emotional judgement.

If AI is repeatedly used to avoid the vulnerability of direct conversation, it may reduce short-term anxiety while making real-life communication feel even more difficult. This is particularly relevant when anger, hurt, fear, or rejection are involved. If that pattern feels familiar, you may find it useful to read about anger in relationships and the hurt beneath arguments.

AI can help someone prepare for a conversation.

It cannot have the conversation for them.

It can suggest language.

It cannot create mutual understanding.

It can reassure someone that their feelings make sense.

It cannot repair a rupture in a real relationship.

That repair requires human contact, emotional risk, and a willingness to be affected by another person.

When Human Support Is Especially Important

AI tools should not be relied upon during mental health emergencies, safeguarding concerns, situations of immediate risk, or severe deterioration.

This is not because AI is always unhelpful. It is because crisis care requires human responsibility, assessment, judgement, and appropriate action.

If you are in immediate danger, feel unable to keep yourself safe, or are worried you may act on suicidal thoughts, seek urgent human support immediately. In the UK, this may include calling 999 in an emergency, going to A&E, contacting NHS 111, or contacting a crisis line such as Samaritans.

AI may generate words that sound supportive.

But crisis support requires more than supportive words.

If you are supporting someone else and feel unsure how to respond, you may find it helpful to read about how to help a loved one who has a mental health issue. If the person is not yet ready for help, the Stages of Change model can also be useful for understanding why pressure and persuasion often backfire.

Therapy in Person, Online, and Between Sessions

AI use often raises a wider question: what kind of support do people need between therapy sessions?

Some people benefit from notes, worksheets, journaling, voice memos, or agreed between-session tasks. Others need help noticing when reflection becomes rumination. For some, online therapy can be convenient and effective, but it still depends on a real therapeutic relationship, not just a digital format.

If you are weighing up different ways of accessing therapy, you may find it useful to read about face-to-face versus remote therapy. The mode of therapy matters less than whether it gives you a safe, thoughtful, clinically informed relationship in which meaningful work can happen.

For readers looking specifically for local support, mental health counselling in Birmingham may be a relevant place to start. For anxiety-related difficulties, Stronger Minds also provides information about anxiety therapy in Birmingham.

The Bigger Picture

Historically, people often feel ashamed of coping strategies before they fully understand them.

AI may become another example of this.

Some people use AI because they are lonely. Some because they are overwhelmed. Some because they need reassurance. Some because they are frightened of burdening others. Some because they struggle to organise emotions internally. Some because predictable responses feel emotionally safer than unpredictable human relationships.

None of this should be dismissed.

But it is also important not to confuse emotional soothing with psychological growth.

In the longer term, psychological adaptability usually depends on the capacity to participate in meaningful human relationships. We need relationships in which we can feel seen, challenged, accepted, repaired with, and emotionally connected. We need social bonds that give us not only reassurance, but belonging.

AI can respond.

But it cannot belong with us.

It can generate words that sound caring, but it cannot create the human bond through which loneliness is often transformed. It can reassure, but it cannot share emotional life with us. It can help someone prepare for a conversation, but it cannot replace the therapeutic effect of having that conversation with another person.

This matters particularly in a world where many people are already struggling with isolation, disconnection, remote working, and reduced everyday social contact. The impact of loneliness and isolation is not only personal; it can also affect work, relationships, and wider wellbeing, as discussed in this Stronger Minds article on mental health, work and loneliness.

That is why the most useful therapeutic question is rarely:

“Should I use AI or not?”

It is more often:

“Is my use of AI helping me move towards a fuller, more connected life — or is it helping me avoid the vulnerability of human connection?”

Therapy does not need to begin from the assumption that AI use is either entirely healthy or entirely unhealthy.

Often, the most useful starting point is simply:

“Let’s try to understand what this is doing for you psychologically.”

That conversation itself may become meaningful therapeutic work.

If this article has raised questions about your own use of AI, anxiety, loneliness, reassurance-seeking, or therapy, you can contact Dr Nick Zygouris at Stronger Minds to discuss whether psychological therapy may be appropriate.

FAQ

Is it unhealthy to use ChatGPT for emotional support?

Not necessarily. Some people use AI tools reflectively for journalling, organising thoughts, or emotional processing. The concern is not simply that AI is being used, but whether it is helping the person move towards reflection, therapy, and real-life connection, or whether it is reinforcing avoidance, dependency, rumination, or compulsive reassurance-seeking.

Can AI replace therapy?

No. AI may offer information, reassurance, or prompts for reflection, but it does not replace psychotherapy. Therapy involves human relationship, clinical judgement, ethical responsibility, emotional attunement, and the possibility of meaningful social bonding. Those relational elements cannot be generated by an AI model.

Should I tell my therapist I use AI?

In most cases, yes. Openness is usually more helpful than secrecy. Telling your therapist allows you to explore what role AI is playing, when you use it, what you seek from it, and whether it is supporting or interfering with your therapy.

Can AI make anxiety or OCD worse?

Potentially, yes. If someone repeatedly uses AI to seek certainty, check whether they are “right”, or neutralise distressing thoughts, it may strengthen reassurance-seeking and compulsive checking patterns. This does not mean all AI use is harmful, but the pattern matters.

Why does talking to AI sometimes feel emotionally comforting?

AI can feel comforting because it is available, responsive, predictable, and non-judgemental. For someone who feels lonely, ashamed, rejected, or overwhelmed, that can be powerful. However, comfort from AI is not the same as the belonging that can come from meaningful human relationships.

What if I feel embarrassed telling my therapist?

The embarrassment itself may be important therapeutic material. It may reflect fears of judgement, dependency, loneliness, shame, or feeling “too much” for other people. You do not need to explain it perfectly. You can begin by saying, “I feel awkward mentioning this, but I have been using AI between sessions.”

Can AI help me prepare for therapy?

Sometimes, yes. AI may help you organise thoughts, identify themes, or prepare what you want to discuss. But it is worth noticing whether preparation helps you become more open in therapy, or whether it becomes another way to seek certainty before speaking honestly.

Key Takeaways

• AI can be used in many different ways. Some people use it to organise thoughts, journal, or prepare for therapy. Others use it for reassurance, certainty, or emotional soothing during distress.
• The central question is not simply whether AI is good or bad. A more useful therapy question is: what is this doing for me psychologically?
• AI may provide reassurance or reflection, but it cannot create genuine human belonging. Meaningful relationships remain central to psychological adaptability, loneliness reduction, and therapeutic change.
• If AI use feels compulsive, secretive, shameful, or as though it is replacing human contact, it may be worth bringing into therapy.
• A thoughtful therapist should be able to explore AI use with curiosity, care, and clinical judgement rather than judgement or defensiveness.

Disclaimer

This article is for educational purposes only and does not constitute medical or psychological advice. AI tools are not substitutes for assessment, psychotherapy, crisis support, or emergency care. If you are experiencing acute distress, suicidal thoughts, or immediate risk, contact emergency services or urgent mental health support services.

Author

Author: Dr Nick Zygouris, Consultant Clinical Psychologist, HCPC-registered and BPS-Chartered
Publish date: 9 May 2026
Last reviewed: 9 May 2026

References

Baier, A. L., Kline, A. C., & Feeny, N. C. (2020). Therapeutic alliance as a mediator of change: A systematic review and evaluation of research. Clinical Psychology Review, 82, 101921. https://doi.org/10.1016/j.cpr.2020.101921

Flückiger, C., Del Re, A. C., Wampold, B. E., & Horvath, A. O. (2018). The alliance in adult psychotherapy: A meta-analytic synthesis. Psychotherapy, 55(4), 316–340. https://doi.org/10.1037/pst0000172

Holt-Lunstad, J. (2024). Social connection as a critical factor for mental and physical health: Evidence, trends, challenges, and future implications. World Psychiatry, 23(3), 312–332. https://doi.org/10.1002/wps.21224

Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). Social relationships and mortality risk: A meta-analytic review. PLoS Medicine, 7(7), e1000316. https://doi.org/10.1371/journal.pmed.1000316

Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science, 10(2), 227–237. https://doi.org/10.1177/1745691614568352

Khawaja, Z., & Bélisle-Pipon, J.-C. (2023). Your robot therapist is not your therapist: Understanding the role of AI-powered mental health chatbots. Frontiers in Digital Health, 5, 1278186. https://doi.org/10.3389/fdgth.2023.1278186

National Health Service. (n.d.). Mental health services. https://www.nhs.uk/mental-health/

National Institute for Health and Care Excellence. (2022). Depression in adults: Treatment and management. NICE guideline NG222. https://www.nice.org.uk/guidance/ng222