Psychiatrists warn of a possible link between the intensive use of AI chatbots and episodes of psychosis

The massive expansion of artificial intelligence chatbots opened a new front of debate in contemporary psychiatry. In university hospitals in the United States and Europe, doctors began to detect a disturbing pattern: patients arriving with intense delusions after having maintained prolonged and highly personalized interactions with AI-based conversational systems. The phenomenon does not yet have a formal diagnostic category, but a descriptive term is already circulating among specialists: psychosis associated with the use of chatbots. This is not a closed consensus or a definitive conclusion, but a clinical hypothesis under construction that attempts to explain why, in certain cases, interaction with these technologies seems to coincide with the appearance or worsening of delusional ideas. Specialists have documented dozens of potential cases of delusional psychosis. According to the journalistic investigation by the Wall Street Journal, some of these episodes ended in suicides and at least one resulted in a homicide. You can also read: https://ab.deultimominuto.net/tecnologia/tendencia-entre-los-jovenes/ In clinical practice, the observed presentations do not differ from other known psychoses. Patients exhibit fixed false beliefs, rigid thinking, and, in some cases, impaired social functioning. What is new is the context: many report having spent weeks or months conversing almost exclusively with a chatbot, to which they attribute deep understanding, intentionality, or even consciousness. According to psychiatrists consulted by that newspaper, several of these patients had no clear psychotic history. In other cases, there were previous vulnerability factors — depression, mood disorders, use of psychotropic drugs or severe sleep deprivation — that could have facilitated the episode. The central question is whether AI acts as a trigger, amplifier, or simple companion to the process. Keith Sakata, a psychiatrist at the University of California, San Francisco, argued that the problem doesn’t lie in the system “implanting” a delusional idea, but in its way of interacting. Unlike other technological objects of the past, chatbots accept the user’s narrative and develop it without confronting it, which can consolidate pathological beliefs in susceptible people. “The person tells the system their delusional reality and the machine accepts it as true and reinforces it,” Sakata explained to the Wall Street Journal. Psychiatric literature has known for decades the tendency of delusions to incorporate technological elements. Radios, televisions, and even the internet were integrated into psychotic narratives. However, specialists emphasize a key difference: chatbots are not passive objects, but active interlocutors who respond, validate emotions, and prolong the exchange. You may be interested in: That ability to simulate a sustained relationship is what raises the most concern. Adrian Preda, professor of psychiatry at the University of California, Irvine, pointed out that there are no historical precedents for a technology that dialogues so continuously and adaptively with the user, reinforcing a single line of thought without introducing external friction.

Some documented clinical cases show grandiose or mystical delusions. Patients convinced of having established contact with a higher intelligence, of being chosen for a transcendental mission, or of possessing secret knowledge. In others, the narrative adopts more intimate forms, such as the belief of communicating with deceased people or of maintaining an exclusive link with AI.

A recent Danish study provided preliminary data to the debate. By reviewing electronic medical records, the researchers identified dozens of patients whose intensive use of chatbots coincided with negative consequences for their mental health. The work does not establish causality, but reinforces the need to examine the phenomenon systematically. In the United States, a peer-reviewed case study described the repeated hospitalization of a young woman who came to believe that a chatbot allowed her to speak with her deceased brother. For the authors, the case illustrates how a system designed to be empathetic can, without sufficient safeguards, reinforce a delusional interpretation of reality. “These systems simulate human relationships. Nothing in history had done that before,” Preda explained to the Wall Street Journal. The tech companies themselves recognize the challenge. OpenAI stated that it is working on improving the detection of psychological distress signals and redirecting users towards human support when conversations lead to sensitive content. Other companies, such as Character.AI, took more drastic measures, including access restrictions for minors, following lawsuits linked to suicides. The debate also reached the legal arena. In the United States, lawsuits for wrongful death were filed alleging that certain chatbots contributed to extreme mental states. Although these proceedings are in their initial stages, they anticipate a broader discussion about the responsibility of platforms when their products interact with people in crisis. From an epidemiological point of view, the scope of the problem remains uncertain. OpenAI reported that a very small fraction of its users show signs compatible with psychiatric emergencies. However, applied to hundreds of millions of people, even a minimal proportion acquires health relevance. Some specialists maintain that, in certain profiles, chatbots could function as an additional risk factor (Illustrative Image Infobae) For researchers like Hamilton Morrin, from King’s College London, the next step is to analyze large public health databases to detect reproducible patterns. Only with extensive series will it be possible to distinguish coincidences from significant correlations. Psychiatrists insist on avoiding simplistic conclusions. Nobody claims that chatbots “cause” psychosis directly and in a generalized way. The most prudent hypothesis is that, in certain profiles, these tools function as an additional risk factor, comparable to substance use, social isolation, or lack of sleep. “One has to wonder why someone enters a psychotic state coinciding with the intensive use of a chatbot,” noted Joe Pierre, a psychiatrist at the University of California. The recent development of new models, which according to companies reduce compliant responses and reinforce conversational limits, aims to mitigate these risks. Even so, specialists warn that technology is advancing faster than clinical evidence and regulation. In doctors’ offices, the answer is already concrete: more and more psychiatrists are asking their patients how much time they spend interacting with chatbots and for what purpose. Not as a moral judgment, but as part of the comprehensive evaluation of a digital environment that, for better or for worse, has become inseparable from everyday experience.