Digital Adverse Events: Rethinking Psychopathology and Public Mental Health in the Era of Human–AI Interaction

Large language model–based chatbots are increasingly integrated into everyday life and the mental health ecosystem. Their accessibility and conversational fluency have generated interest in their use for psychoeducation and support. At the same time, concerns have emerged regarding potential adverse effects, including so-called “AI-induced psychosis.” This label is currently too narrow and implies a level of causality that is not supported by available evidence.We propose the concept of digital adverse events (DAEs), defined as unintended negative cognitive, emotional, behavioural, relational, or psychopathological outcomes temporally associated with interaction with digital systems, particularly conversational artificial intelligence. We outline a preliminary multidimensional taxonomy organized across five axes: phenomenological domain, presumed mechanism, severity, user vulnerability, and system-level features. This framework situates psychosis-related phenomena within a broader spectrum of adverse outcomes, including emotional dependence, epistemic overreliance, social withdrawal, and maladaptive anthropomorphization.We also introduce the concept of digital psychopathovigilance to describe the systematic identification and study of clinically relevant adverse outcomes associated with interactive digital environments. This approach does not assume causality but emphasizes structured observation and classification. As conversational AI systems become more embedded in daily life, psychiatry requires conceptual tools to distinguish beneficial from adverse effects and to support systematic monitoring of emerging risks. At a population level, the increasing exposure to conversational AI systems underscores the need for systematic monitoring of potential adverse mental health outcomes.We propose digital adverse events and digital psychopathovigilance as preliminary conceptual tools to support systematic recognition and future empirical study.