Guardian Science Weekly

‘AI psychosis’: could chatbots fuel delusional thinking?

Loading player...
There are increasing reports of people experiencing delusions after intensive use of AI chatbots. The phenomenon, dubbed ‘AI psychosis’, has raised concerns that features built into large language models may contribute to some users losing touch with reality. Madeleine Finlay speaks to Dr Hamilton Morrin, a psychiatrist and researcher at King’s College London, about his recent preprint exploring who is at risk and how models could be made safer. Help support our independent journalism at theguardian.com/sciencepod
28 Aug 2025 English United Kingdom Science · Nature

Other recent episodes

Everything you need to know about Artemis II so far

This week Artemis II’s four-astronaut crew broke Apollo 13’s distance record, becoming the humans to travel the farthest from Earth. Now on their way home, the team has experienced tech malfunctions, views like no other and moments of intense emotion, all in under 10 days. To find out about all…
9 Apr 19 min

Can we eradicate a second human disease?

The number of human cases of guinea worm, a painful and debilitating tropical illness, fell to a record low of just 10 last year, according to the Carter Centre, the foundation set up by the late former US president Jimmy Carter. But despite years-long declines, it remains almost impossible to…
7 Apr 15 min

Sauna and cold plunge: where does the evidence stand?

Saunas and cold plunge pools are popping up everywhere in the UK, bringing fiery heat and icy cold to a beach, city farm or park near you. Their users will be ready with all the reasons why it’s good for both the mind and the body. But what’s the evidence…
2 Apr 16 min
All episodes