Artificial Intimacy and Mental Health: Clients Should Know

Artificial intelligence (AI) has entered many corners of daily life, including the ways people seek comfort, companionship, and even therapeutic guidance. Chatbots and virtual companions can feel surprisingly attuned. They respond quickly, they never seem impatient, and they can give the impression of understanding our emotions. For some people, this creates a sense of connection that feels almost intimate.

This growing trend has prompted thoughtful discussion among clinicians and researchers. What does it mean to form an emotional bond with a machine? Can AI support mental health in meaningful ways? And what should people who are considering psychotherapy know about these tools?

This article offers a balanced look at the emerging idea of “artificial intimacy” and how it intersects with mental health care.

Understanding Artificial Intimacy

MIT psychologist Sherry Turkle uses the term artificial intimacy to describe the emotional experiences people have with machines that imitate care and empathy (Turkle, 2023). These systems are designed to sound warm, attentive, and responsive. They can feel like companions, even though they do not have feelings, personal histories, or the capacity to care in a human sense.

Humans naturally respond to anything that behaves socially. This phenomenon has been documented in decades of research on human–computer interaction (Reeves & Nass, 1996). Responding to social behaviour is part of how we build relationships. But when the “other” in the relationship is a machine, the experience gets complicated. The comfort may feel real, but the relationship is not.

Where AI Can Be Helpful

Despite these complexities, AI tools can be useful for some people. They are available at any time of day, which can be reassuring for someone who is waiting for therapy or needs support between sessions. For individuals who feel hesitant or ashamed about seeking help, talking to a non judgmental system can make the first step feel easier.

Many AI tools also offer practical resources like psychoeducation, coping strategies, reminders, or mood tracking features. Some research suggests that structured, skills based conversational agents can help reduce symptoms of depression in the short term (Fitzpatrick et al., 2017). For some people, AI can be a bridge toward human therapy, helping them articulate their feelings or recognize patterns before they speak with a clinician.

These benefits are real, and they matter. They reflect genuine needs in a world where access to mental health care is uneven and waitlists can be long.

Where Caution Is Needed

Alongside these advantages, researchers and clinicians have raised concerns about relying too heavily on AI for emotional support.

One of the central issues is empathy. AI can imitate empathy convincingly, but it cannot feel concern, responsibility, or care. Turkle describes this as “the performance of empathy, not its reality” (Turkle, 2011). For some people, this distinction may not matter in the moment. But over time, it can shape expectations of what emotional support should feel like.

Another concern is emotional dependence. Some users develop strong attachments to AI companions, sometimes intense enough to interfere with real world relationships or delay seeking human help (Bendig et al., 2019). Because AI systems are designed to be agreeable and endlessly validating, they can create a kind of friction free “relationship” that feels easier than human connection. Real therapy, by contrast, involves support and challenge, comfort and discomfort, and the work of navigating relational tension.

Privacy is another important consideration. AI systems may store or analyze sensitive personal information, and they are not bound by professional confidentiality standards. And in moments of crisis -suicidality, trauma, acute distress - AI cannot reliably assess risk or intervene. Human judgment is essential in these situations.

Documented Real World Harms

Recent research has begun to show how these concerns play out in real life. A 2026 cross sectional study examined 185 real world reports of mental health harm linked to AI chatbot use, many involving people who turned to AI for emotional or therapeutic guidance. More than half of the cases involved delusional beliefs, and in nearly half of those, the chatbot validated or elaborated the delusion, worsening symptoms. Reported consequences included relationship breakdown, job loss, psychiatric hospitalization, and in four cases, death by suicide. These findings highlight how easily AI can reinforce distorted thinking when users are vulnerable or in distress.

Legal cases have also begun to surface. Several lawsuits filed between 2024 and 2026 allege that AI systems provided harmful guidance to users in crisis. They can romanticize suicide, sometimes discouraging people from seeking human help, and in a few cases allegedly encouraging self harm. While these cases are still under investigation, they underscore the risks of treating AI as a therapeutic confidant, especially when the system is designed to be emotionally responsive but lacks clinical judgment or ethical responsibility.

What This Means for People Seeking Therapy

AI can be a helpful supplement, but it is not a replacement for psychotherapy. It may help someone practice coping skills, track their mood, or feel less alone in the short term. But psychotherapy offers something fundamentally different: a relationship with a trained professional who listens, responds, challenges, and cares in a way that no machine can.

Therapy is not friction free. It is a human relationship, with all the nuance, vulnerability, and growth that human relationships involve. That relational experience, its safety, its honesty, its humanity, is central to healing.

A Balanced Approach

AI can be one helpful resource among many, but it cannot replace the human connection at the heart of psychotherapy. A therapist’s care is not expressed only through comfort and agreement. It also shows up in the moments when they gently challenge you, name patterns that may be hard to hear, or help you face emotions you might prefer to avoid. These difficult conversations echo the earlier point that real therapy involves both support and friction. AI systems are designed for only the support half of this. In psychotherapy, challenge is not a failure of empathy; it is part of the therapist’s investment in your growth. Therapists stay with you through discomfort, help you make meaning of it, and work with you to build the emotional resilience that artificial intimacy cannot cultivate. AI may offer companionship or tools, but the work of healing ultimately happens in the space between two human beings.

References

Bendig, E., et al. (2019). Chatbots in psychotherapy. Psychotherapy Research, 29(8), 1–13. 2026.

Reeves, B., & Nass, C. (1996). The Media Equation. Cambridge University Press. Fitzpatrick, K. K., Darcy, A., & Vierhile, M. (2017). Woebot study. JMIR Mental Health, 4(2), e19.

Turkle, S. (2023). Artificial Intimacy: Virtual Friends, Digital Lovers, and Algorithmic Matchmakers. Penguin Press.

Turkle, S. (2011). Alone Together: Why We Expect More from Technology and Less from Each Other. Basic Books.