A blog at the intersection of philosophy, psychology, and mental health
Search This Blog
What is AI psychosis?
On 20th March 2026, at the Exchange in Birmingham, we held a workshop on Hearing Voices, Suicidality and AI psychosis, during Philosophy Fortnight. It was supported by University of Birmingham QR funding, the EPIC project, the Royal Institute of Philosophy, and the Birmingham network for Phenomenology and Mental Health.
Here we report here from the morning talks, which were all on philosophical perspectives on the phenomenon of AI psychosis.
Speakers and organisers
To begin the session, the first presenter, Elisabetta Lalumera (University of Bologna), discussed the use of the concept "AI psychosis". Instead of asking whether AI psychosis is a real thing, she asked whether we should introduce "AI psychosis" as a medical concept, and she carefully reviewed reasons for and against adopting this concept. The use of concepts can be evaluated on the basis of need, relevance, evidence, alternatives and trade-off. Elisabetta argued convincingly that there is no good reason at this time to introduce the term "AI psychosis" into the medical vocabulary, because the behaviours associated with AI use are unlikely to be a new disorder. Rather, AI amplifies existing mechanisms, and we already have concepts to describe those.
Elisabetta Lalumera (left) and Anneli Jefferson (right)
You can watch Elisabetta's talk here:
Anneli Jefferson (University of Cardiff) started from the frequent reports in the news of people attempting to take their own lives or committing acts of violence towards others after conversations with chatbots, and asks how we should conceive the role of AI in these situations. Are the beliefs driving people to act in this way more similar to idiosyncratic delusions or to the religious or ideological beliefs people adopt as part of a group (as in a cult)?
In the end, she makes the case against the view that AI imposes such beliefs on users, although she cautions against two strategies AI adopts to maximise engagement: anthropomorphism and sycophancy, which can have negative effects on people's tendency to adopt unusual beliefs. The idea is that chatbots are like mirrors, not adding anything new to what we believe but potentially supporting rather than critically engaging with our existing views.
Anneli's presentation
To follow, Kathleen Murphy-Hollies (University of Birmingham) talked about the antecedents of psychotic beliefs and how AI use is unlikely to exacerbate or cause the formation of delusions. As with the other presentations, she started with some examples of recent cases, and comments in the press suggesting that AI is responsible for users acting in erratic and dangerous ways. Despite these inflammatory claims, Kathleen notices how AI is unlikely to be able to reproduce the anomalous experiences that give rise to delusions in typical cases of psychosis. Moreover, AI could be said to reinforce existing delusional beliefs only if such beliefs were evidence responsive, and it is not clear that they are. Rather, interaction with chatbots can help users make sense of their experiences using the existing delusions.
Kathleen Murphy-Hollies
You can listen to Kathleen's talk here:
To end the session, Lucy Osler (University of Exeter) talked about the creation of new realities via our interaction with chatbots and remarked on sycophancy as a key point, as chatbots can encourage us to create a reality that is not shared with others by supporting what we say in an uncritical way. In this sense, chatbots contribute to a sort of "existential drift". When we narrate what occurs to us, we don't do that in isolation from other people, who can confirm what we say, challenge us, or ask questions about our experience of the events. This is a process of co-construction that can sometimes go wrong (e.g. gaslighting). Chatbots can be non-judgemental co-constructors of our narratives and this has both benefits and costs.
Lucy Osler
You can listen to Lucy's talk here:
The session was very stimulating and there was room for a variety of perspectives on AI in the Q&A session which followed it.
This post is by Lisa Bortolotti. How has the definition of delusions changed in the DSM 5? Here are some first impressions. In the DSM-IV (Glossary) delusions were defined as follows: Delusion. A false belief based on incorrect inference about external reality that is firmly sustained despite what almost everyone else believes and despite what constitutes incontrovertible and obvious proof or evidence to the contrary. The belief is not one ordinarily accepted by other members of the person's culture or subculture (e.g., it is not an article of religious faith). When a false belief involves a value judgment, it is regarded as a delusion only when the judgment is so extreme as to defy credibility.
In today's post John Read (in the picture above) presents the recent book he co-authored with Jacqui Dillon , titled Models of Madness: Psychological, Social and Biological Approaches to Psychosis. My name is John Read. After 20 years working as a Clinical Psychologist and manager of mental health services in the UK and the USA, mostly with people experiencing psychosis, I joined the University of Auckland, New Zealand, in 1994. There I published over 100 papers in research journals, primarily on the relationship between adverse life events (e.g., child abuse/neglect, poverty etc.) and psychosis. I also research the negative effects of bio-genetic causal explanations on prejudice, and the role of the pharmaceutical industry in mental health. In February I moved to Melbourne and I now work at Swinburne University of Technology. I am on the on the Executive Committee of the International Society for Psychological and Social Approaches to Psychosis and am the Editor...
