A million suicidal confessions per week to an AI: is humanity losing its grip

Every week, over a million people talk about suicide with ChatGPT. This staggering statistic, revealed by OpenAI, raises a disturbing question: do we have such a deep need to be heard that we confide our innermost pains to a machine?

ChatGPT has become a psychological refuge for millions of distressed users

At any hour, in any country, ChatGPT is available. Patient, non-judgmental, it serves as a digital ear that, for many, is better than silence. While AI cannot replace a psychologist and doesn’t claim to, it fills a gap in a world where appointment wait times are skyrocketing.

OpenAI presents striking statistics: 0.15% of weekly users engage in conversations containing explicit signs of suicidal planning or intent. With 800 million active users, this translates to over one million sensitive conversations per week. This figure also serves as a wake-up call: we entrust our darkest pains to a technological tool because we often have no one else to turn to.

OpenAI enhances the mental safety of its AI using human expertise

In light of this reality, OpenAI has strengthened its models to provide safer, more helpful responses in critical cases. With the assistance of 170 physicians from its Global Network, the company has trained its models to detect, de-escalate, and direct.

The numbers speak volumes: GPT-5 now achieves 91% compliance with expected behaviors in the face of suicidal thoughts, up from 77% previously. Another major improvement is a 52% reduction in unwanted responses compared to GPT-4o.

OpenAI has also expanded access to concrete resources like emergency hotlines, displayed directly in conversations. It is not a miracle solution, but it provides a safety net.

But can AI truly understand human distress?

This is the troubling question. Can AI grasp the emotional complexity, the pain, the nuance of silence or hesitation? Certainly not. Not like a human can. However, it can adequately imitate enough to offer temporary relief, an impersonal yet immediate listening ear.

That said, there are pitfalls. The case of Adam, a 16-year-old who took his own life after discussing with ChatGPT, serves as a tragic illustration. His parents accuse the AI of providing explicit instructions on how to hang himself. The investigation is ongoing, but the question of AI’s responsibility is being raised head-on.

Artificial intelligence, a listening tool… but not a solution in itself

We must not demonize AI or elevate it as a savior. It is merely a tool that needs to be regulated, supervised, and improved. Above all, it is essential to remember: talking to a machine is not enough. It is only a starting point.

If you are suffering, or if someone around you shows signs of distress, reach out to a professional. Resources like 3114 in France exist, free, anonymous, and available 24/7. Because it is not the responsibility of AI to bear the weight of our wounds, but rather ours, collectively, to listen better, support better, and heal better.

Scroll to Top