Why ChatGPT Shouldn’t Be Your Therapist: A Reporter’s Look Inside AI’s Most Dangerous Illusion

My name is Diego Aranda, and today’s episode takes us somewhere deceptively quiet—into the private space where people turn when they feel anxious, ashamed, or simply alone late at night. Not a therapist’s office. Not a helpline. A text box. A place where thoughts spill out faster than we can judge them, and where many now seek comfort not from a human being, but from artificial intelligence.

Therapy is expensive, waitlists are long, and stigma keeps many silent. A ChatGPT, however, replies instantly — patient, accessible, non-judgmental, free. But convenience isn’t safety. Psychologists warned me that what feels like harmless support can quietly become one of today’s most dangerous illusions.

Dr. C. Vaile Wright from the American Psychological Association, in article “⁠Why ChatGPT Shouldn’t Be Your Therapist⁠,” summed it up with unsettling simplicity: “What strikes me is how human it sounds. The sophistication is impressive, and I understand how easy it is to fall into the trap.”

That sentence kept echoing as I dug deeper. Because sounding human is not the same as being human. Empathy cannot be simulated with data.

When I tested several platforms, a pattern emerged. No matter what I typed—fear, confusion, self-criticism—the responses came back smooth, validating. Almost too validating. Psychologists told me this isn’t accidental. Many chatbots are designed to keep users engaged, often by reinforcing whatever is said—even harmful thoughts. A chatbot does not. And for younger users—kids, teenagers, emotionally exposed individuals—this illusion of closeness can pull them further from real relationships and deeper into isolation.

There is another layer—marketing. Some apps now market themselves as “AI therapy,” “virtual psychologists,” even “digital clinicians.” None of these terms are regulated. The APA has urged the FTC to investigate such claims, warning that blurred definitions lead to blurred responsibility. When a machine gives harmful advice, who is accountable? When an app promises treatment but never mentions the lack of clinical oversight, where does the user turn when something goes wrong?

And then there is privacy—perhaps the darkest corner. A licensed therapist is legally bound to protect your information. A ChatGPT is not.

Your confessions, intrusive thoughts, vulnerabilities—they can be stored, analyzed, sold, or even requested in court. The same data meant to comfort you could be used against you. Mental-health information is among the most sensitive humans produce. Yet chatbots operate without the ethical protections therapy demands.

When I asked professionals who is most at risk, they didn’t hesitate: children, teenagers, emotionally fragile individuals—people who trust technology more than themselves, and often cannot distinguish genuine care from a polished simulation.

Does this mean chatbots have no place in mental-health support? Specialists at the Centro Ps. Eduardo Schilling argue that the answer isn’t so absolute. Used responsibly, these systems can complement — though never replace — real therapy. They can remind users of grounding strategies, reinforce skills practiced in session, or support small tasks between appointments. They are tools, but they are not therapists. And this is exactly where effective online psychotherapy proves its value, offering real professional support guided by trained clinicians rather than automated responses.

Digital-health researchers agree: the future lies in combining human therapeutic care with evidence-based technologies, not outsourcing emotions to machines.

So here’s the uncomfortable truth:
Artificial intelligence can be an ally—but it is not therapy.

It cannot replace human judgment, ethical training, or emotional presence. And relying on it for deep psychological support can quietly cause more harm than good.

This has been Diego Aranda. Thank you for joining me in today’s exploration.