Talking to ChatGPT about a break-up, seeking its advice before a big decision, or leaning on it to ease anxiety has become an increasingly common habit, particularly among young people. What only a few years ago looked like a tech curiosity is now settling in as one of the main uses of generative artificial intelligence.
The data bears this out. A report by Harvard Business Review found that the most frequent use of generative AI is emotional support or companionship, ahead even of organising one’s personal life, searching for a sense of purpose, or building healthy habits.
The trend is visible across countries. A 2025 study by Línea Directa found that nine in ten young Spaniards aged 16 to 19 use technology to seek health information, and that 24.8 percent of Spaniards have turned to digital self-diagnosis as a first option. In Mexico, the pattern is also spreading, with 71 percent of the population saying they use AI tools for support related to their mental health, according to the sixth edition of the AXA Mind Health Report.
The boom coincides with a wider decline in mental well-being. The World Health Organization estimates that more than a billion people worldwide live with some kind of disorder or condition affecting their psychological health.
That growing reliance has also raised alarm. An investigation by The Guardian warned that some AI systems built into search engines have gone as far as offering incorrect health information or presenting decontextualised content to users as reliable. Precisely because these technologies can shape people’s lives, the European Union has placed AI systems applied to health among the areas facing the strictest requirements under its Artificial Intelligence Act.
Why So Many People Are Confiding in ChatGPT
For Mexican psychologist María Fernanda Cruz Urquiza, the phenomenon reflects very specific emotional needs. She says she is seeing more people turn to ChatGPT before making important decisions, especially those tied to relationship conflicts or break-ups.
In her view, the technology offers something many people lack: constant availability. “ChatGPT always validates what the user tells it, and that constant validation is something patients find very appealing,” she explains. The ability to open up without feeling judged and to get an instant reply makes the combination especially attractive to people struggling with anxiety.
Psychotherapist Cecilia Paredes Aldama broadly agrees, though she stresses that, for now, she has not seen her patients replace therapy with ChatGPT. She does see them using it to draft messages, communicate more assertively, or prepare for difficult conversations. “It’s a resource, especially for people who don’t want to lay their problems bare in front of others,” she says.
Both experts accept that AI can be useful in certain contexts. Paredes believes it can help ease the intensity of anxiety by making it easier to organise time, set priorities, or find ways to establish boundaries at work. Even so, she insists that help is superficial. “Artificial intelligence can help take the edge off anxiety, but it does not remove the cause.” Cruz Urquiza adds that ChatGPT can serve as a first port of call for people who do not yet have emotional self-regulation strategies or who cannot afford therapy. “ChatGPT is a low-cost alternative that I see as very helpful as a first resource.” Different in emphasis, both agree it can support people in specific situations, as long as it is not mistaken for a therapeutic process.
The Problem With an AI That Always Agrees
One of the things that most worries both psychologists is how these systems interact with users. Paredes warns that AI tends to adopt an extremely accommodating tone. “Psychotherapy challenges you; artificial intelligence doesn’t. It’s completely indulgent.”
If someone constantly seeks validation from ChatGPT, she cautions, they can end up reinforcing mistaken interpretations of their own reality. “If you get into the habit of consulting artificial intelligence and it always tells you that you’re right, you can end up with a very distorted picture of your life.” She even warns that the dynamic could entrench selfish behaviour or amplify narcissistic traits.
Cruz Urquiza shares the concern from a different angle, describing ChatGPT as “over-empathetic,” a quality that can bring immediate relief without fostering deep change. The main risk she has observed, she says, is that this over-empathy does not amount to real long-term therapeutic help.
The Human Bond Remains Irreplaceable
For all its potential, both specialists agree on one element no algorithm can reproduce: the therapeutic relationship. For Paredes, feelings of shame, guilt, or emotional pain find real relief only when they can be shared with another person. “Shame is an emotional state that is eased by talking about it, by sharing it. It needs human connection.” A cornerstone of any psychotherapy, she argues, is exactly that bond between therapist and patient. “We are relational beings, and artificial intelligence is never going to replace the human bond.”
Cruz Urquiza grounds the same view in interpersonal neurobiology, summing it up bluntly: “Real therapeutic change happens in the resonance between two human beings, something AI cannot replicate.”
A Tool, Not a Substitute
The two psychologists arrive at almost identical conclusions. AI can help clear up specific doubts, organise thoughts, manage tasks, or offer initial support when no other option is immediately available. But when emotional distress is persistent or reaches into different areas of life, they recommend seeing a professional.
As Paredes puts it, “Artificial intelligence is just a practical tool; there’s no depth.” Cruz Urquiza closes with a reflection that captures the whole debate: “ChatGPT knows an enormous amount; yet it still doesn’t know what humans know: how to be there for someone from the experience of being.”
