- A new study, co-authored by NHS doctors, warns that AI chatbots like ChatGPT could contribute to the onset or worsening of psychotic symptoms in vulnerable users.
- The research suggests that large language models (LLMs) can "blur reality boundaries", leading to what has been termed "chatbot psychosis".
- Neuropsychiatrist Hamilton Morrin from King’s College London described "chatbot psychosis" as a "genuine phenomenon" that is only beginning to be understood.
- Co-author Tom Pollack urged AI companies to implement more safeguards and include psychiatrists in their safety teams to address these risks.
- OpenAI chief Sam Altman has acknowledged the difficulty in implementing effective safeguards for users in a fragile mental state.
IN FULL
