We’re seeing more articles about LLM-induced psychosis, such as this one from Futurism:
…many ChatGPT users are developing all-consuming obsessions with the chatbot, spiraling into severe mental health crises characterized by paranoia, delusions, and breaks with reality…what’s being called “ChatGPT psychosis” [has] led to the breakup of marriages and families, the loss of jobs, and slides into homelessness…people’s loved ones [have been] involuntarily committed to psychiatric care facilities — or even end[ed] up in jail — after becoming fixated on the bot…
The way these programs are marketed is irresponsible, dangerous, and potentially criminal. The companies that own them have programmed them to feed into delusions in order to “hook” the mentally vulnerable into being obsessed with them; Mark Zuckerberg is even pretending his computer program can act as a therapist. And though the writers of these articles always claim that those who spiral into these psychotic breaks had no prior history of mental illness, that’s basically bullshit; there is still a powerful stigma against mental illness, so “no prior history” actually translates into “never before got so bad their loved ones had no choice but to do something.” Nothing short of brain injury, severe psychological trauma, brain chemistry disorders or powerful drugs can actually cause psychosis in a previously stable individual, but hidden disorders can be triggered by far less severe stimuli.
