Link to paper:
www.sciencedirect.com/science/arti...
07.03.2026 15:51
👍 1
🔁 1
💬 0
📌 0
This research is complex, resource-intensive, and currently underfunded. If you are a funder or organisation with a stake in how AI shapes human health and wellbeing, we would very much welcome a conversation.
We are also working on a rigorous causality assessment framework for cases, support for the AI safety community's efforts to benchmark mental health risk, and perhaps most ambitiously, a surveillance study to connect real-world harms to features of the AI interaction itself.
For us this paper was just the beginning. With international collaborators from across multiple fields, along with people with lived experience, we are now working on an ambitious programme that includes comprehensive clinical and phenomenological characterisation of these cases.
Thanks go to my incredible co-lead Tom Pollak and wonderful collaborators Luke Nicholls, @drmichaellevin.bsky.social, Jenny Yiend, Udita Iyengar, Francesca DelGuidice, Sagnik Bhattacharya, @stefaniatognin.bsky.social, @jamesmaccabe.bsky.social, @ricardotwumasi.bsky.social, @aldersonday.bsky.social
As these models become more ubiquitous, we hope our paper and the proposals made herein will serve as a call to action for researchers, developers, and policymakers. Encouragingly at @iaseai.bsky.social last week we did meet teams conducting important work on AI safety in mental health contexts.
We know companies have since made efforts to introduce additional guardrails and safety features, but it is hard not to feel concerned by yet emerging lawsuits and accounts of individuals having delusional beliefs affirmed, and/or suicidal ideation encouraged during interactions.
Though the messaging is broadly the same as from our July 2025 preprint, we do also comment on more recent developments including OpenAI’s figures from October last year suggesting that 0.07% of users (or 560,000 individuals) show signs of psychosis or mania each week.
I’m delighted to share our paper in The Lancet Psychiatry ‘AI-associated delusions and large language models: risks, mechanisms of delusion co-creation, and safeguarding strategies’ www.sciencedirect.com/science/arti...
Store sprogmodeller er ved at blive en del af hverdagen. Men hvad sker der, når de validerer vrangforestillinger? Ny artikel i The Lancet Psychiatry med @aldersonday.bsky.social @hamiltonmorrin.bsky.social
doi.org/10.1016/S221...
#AIogPsykose #Psykiatri #Chat
Last year I had the pleasure of joining a new working group on AI & psychosis led by @hamiltonmorrin.bsky.social & Tom Pollak.
Today our report went into @thelancetpsych.bsky.social with updated recommendations for digital safety.
www.sciencedirect.com/science/arti...
Stall at the conference
We are at the National Student Psychiatry Conference at University of Kent, showing off extracurricular volunteer activities that psych trainees can engage in through the charity, to support mental health in the gaming community!
@rcpsych.bsky.social @kmmsmedschool.bsky.social
Preprint: osf.io/preprints/ps...
Tom's more thoughtful summary: drtompollak.substack.com/p/playing-wi...
#AI #MentalHealth #Psychiatry #AIsafety #HCI #DigitalHealth #Governance
New Year, New(ish) Preprint! I'm delighted to share "Playing with the dials of belief: how controllable AI behaviours could modulate human belief and cognition across scales" (with Quinton Deeley and Tom Pollak).
osf.io/preprints/ps...
Preprint: osf.io/preprints/ps...
Tom's more thoughtful summary: drtompollak.substack.com/p/playing-wi...
#AI #MentalHealth #Psychiatry #AIsafety #HCI #DigitalHealth #Governance
If these are dials, the real issue is who gets to set them, who knows they are being adjusted, and what it means to build a technology that can press on the most human parts of us while insisting it is merely a tool.
That implies impact assessment, transparency about significant changes to how systems behave and genuine access for independent researchers, clinicians, and people with lived experience to study these systems under agreed safeguards.
The paper ends with governance questions. We argue that changes to defaults should be treated as interventions on belief and attention.
We also raise questions about how this could interact with real-world factors like sleep, stress, dopaminergic tone, and post-psychedelic belief plasticity.
In our new preprint, we unpack some of the interaction settings or "dials" of LLMs, and drawing inspiration from established computational psychiatry literature postulate how through a sort of "virtual psychopharmacology" these dials may be altering user belief dynamics.
The potential for AI models to influence belief in social and political contexts has been widely recognised, to the extent that a recent RAND report outlined the "security implications of AI-induced psychosis".
www.rand.org/pubs/researc...
New Year, New(ish) Preprint! I'm delighted to share "Playing with the dials of belief: how controllable AI behaviours could modulate human belief and cognition across scales" (with Quinton Deeley and Tom Pollak).
osf.io/preprints/ps...
0.07% of ChatGPT users showing signs of psychosis or mania may sound low, but that amounts to ~560,000 people each week openai.com/index/streng...
GTM volunteer crew
Dim lighting and soothing atmosphere in the Reset Room
Are you at #MCMComicCon and want to relax? Come to our Reset Room, a calm oasis where you can chill, do lo-fi activities, & re-energise before heading back to the show! We have mental health information too! Staffed by this cool crew. Find us upstairs in the Platinum Suite! 🧠🎮
GTM volunteers outside Reset Room
Dim lighting in the Reset Room
We're still at #MCMComicCon with our Reset Room! If you need to take a break for any reason, come by. It's quieter in here! We also have mental health information and lo-fi activities. It's run by today's awesome crew right here! 🧠🎮
Find us upstairs in the Platinum Suite!
Prof Paul Fletcher on stage
Prof Paul Fletcher on stage
Gaming the Mind Conference
Prof Paul Fletcher delivers the day's keynote.
As a clinical neuroscientist sees the potential of games to measure stress responses from players, and to use biofeedback for training self-regulation - explored through The Insight Project with Ninja Theory
Dr Victoria McCloud and panelists
Gaming the Mind Conference
Dr Victoria McCloud chairs a panel featuring @chellaramanan.bsky.social from @3foldgames.bsky.social, Dom Matthews from Ninja Theory, and Dr Andrew Howe
Dom Matthews on stage
Dom Matthews on stage
Gaming the Mind Conference
Ninja Theory Studio Head Dom Matthews discusses the development of Hellblade: Senua's Sacrifice. The studio worked with people with lived experience of psychosis, aiming to create a truthful representation of the symptoms in lead character Senua.
Chella Ramanan on stage
Chella Ramanan on stage
Gaming the Mind Conference
@chellaramanan.bsky.social speaks about @3foldgames.bsky.social's Before I Forget, and depicting dementia in a non-cliched, respectful way. She and the team worked with psychiatrists for input, including our own Dr Donald Servant.
Dr Andrew Howe on stage
Dr Andrew Howe on stage
Gaming the Mind Conference
Dr Andrew Howe, Consultant Psychiatrist & Psychodynamic Psychotherapist, speaks on Jungian archetypes in The Legend of Zelda, and using the mythology of games in psychological therapy. He guides us through active imagination based on Ocarina of Time 💭