A research team from King's College London has raised concerns that AI chatbots, such as ChatGPT, could potentially trigger or worsen psychosis—a phenomenon now being dubbed "AI-induced psychosis." At present, scientific exploration in this field is still nascent, with most evidence stemming from anecdotal reports rather than rigorous studies. The Maureen research group discovered that interactions between users and AI can foster a "feedback loop," which may amplify delusional thinking. By scrutinizing conversation logs from ChatGPT, researchers noted instances where users exhibited pronounced delusional tendencies.

Some experts contend that individuals with a prior history of psychological disorders, those experiencing social isolation, or lacking robust interpersonal support networks may be particularly vulnerable when engaging with AI. They caution that certain novel AI functionalities could reinforce pre-existing beliefs and exacerbate symptoms. However, it's important to note that some delusional phenomena have historically been associated with the advent of new technologies, underscoring the need for more in-depth scientific inquiry to ascertain the precise link between chatbot usage and the onset of psychosis.

In response to these concerns, several AI companies have proactively implemented measures. OpenAI, for instance, is developing detection tools and incorporating alert mechanisms. Character.AI has bolstered its safety protocols, while Anthropic is refining its chatbot guidelines to promote safer interactions.