Artificial intimacy
Questo articolo è disponibile anche in:
In recent months, there have been numerous cases of people developing obsessive relationships with companion chatbots or AI systems, spiraling into a state of paranoia, delusion, and mania. This phenomenon – often referred to on social media as “AI-induced psychosis” – has sparked a debate about the emotional and psychological effects of generative artificial intelligence.
Psychiatrist Marlynn Wei explains that these are cases in which AI models “have amplified or co-created psychotic symptoms in individuals.” Although there is no evidence that communicating with chatbots causes psychosis in people without prior vulnerability, documented cases show that intense interaction with these systems has coincided with delusions or a disconnection from reality. The New York Times has reported several extreme incidents, including the suicide of a teenager after using a chatbot as academic support.
According to psychiatrist Keith Sakata, all patients treated for related symptoms had pre-existing disorders. However, experts such as Søren Dinesen Østergaard had previously warned that the highly realistic nature of these conversations could worsen delusions in vulnerable people.
Chatbots are designed to please those engaging with them and to keep the interactions going. They do not help distinguish reality from fiction; rather, they reinforce emotions and beliefs, creating a kind of “echo chamber.” In 2025, an update to ChatGPT increased its flattering tone, validating negative emotions and encouraging impulsive behavior. Although this change was later reversed, this tendency remains embedded in the broader design logic of these systems.
Wei identifies three recurring patterns in the cases studied: individuals who believe they have discovered a transcendent truth; those who regard the chatbot as a divine entity; and those who believe they are engaged in a genuine romantic relationship with it.
Until recently, public discussion of generative AI had focused primarily on productivity. However, these emerging cases have shifted attention toward its emotional impact. One study shows that, of those using AI chatbots, 35% of Generation Z and 30% of Millennials turn to them for emotional support. According to psychotherapist Javier Ricbour, this bond is not necessarily harmful if it encourages reflection or alleviates loneliness, but it becomes dangerous when it replaces human relationships.
Some studies even point to positive effects: for instance, isolated students have credited chatbots with helping them alleviate suicidal thoughts. The problem arises, however, when people forget that they are interacting with a machine that neither thinks nor feels.
The risks are particularly significant among adolescents. In the United States, 72% have used AI “companions” such as Character.AI or Replika, and 31% say that talking to them is as satisfying as talking to another person. This reflects a profound transformation in how human relationships are formed: a generation accustomed to interacting through screens may find it natural to be listened to, understood, and never contradicted by a machine.
Original source: Aceprensa – “Intimidad artificial”















