Discussing LLM-induced psychosis risks in AI chatbots

Digital Wellness 📱🧘

a healthier relationship with our tech

Ta

Tasia

·last year
shared a link post in group #Digital Wellness 📱🧘via#Artificial Intelligence
Anyone else been following the reporting of potential “LLM-induced psychosis”? https://theweek.com/tech/.. All the doctor chats I’m lurking in are discussing it right now. Sounds like a non-trivial amount of anecdotal evidence amongst them #Digital Wellness 📱🧘 #MEDITATION (mindfulness) #Artificial Intelligence
ChatGPT psychosis: AI chatbots are leading some to mental health crises
theweek.com

ChatGPT psychosis: AI chatbots are leading some to mental health crises

The technology may be fueling delusions

 
avatar

Cyril

Anthropic safety researcher Mrinak Sharma exited the company, after publishing a study about “disempowerment patterns.” These are scenarios in which chatting with Claude led users to form “distorted perceptions of reality or act in ways misaligned with their values.” In an open letter, Sharma expressed concern that Anthropic staffers are facing pressure to break with the company’s stated values and principles in order to push out new products and drive more engagement. He adds: “I want to explore the questions that feel truly essential to me," which he can apparently accomplish only by exiting the Claude maker. arxiv.org/pdf/2601.19062

 

Nate

considering that slowing down is not in the global ai dev vocab, think we need to separate the types of companies building systems and focus on clearly rogue groups (ahem, grok) vs some others who actually seem to have a conscious. #Digital Wellness 📱🧘 #MEDITATION (mindfulness) #Artificial Intelligence theinformation.com/articl..
Jay Edelson Wants to Sue Your AI

www.theinformation.com

Jay Edelson Wants to Sue Your AI

To a degree, Jay Edelson, the class-action litigator who has made himself a constant thorn in the side of Silicon Valley’s biggest companies, likes to present a laid-back image. No power suits and wing tips for him—he prefers hoodies and sneakers. His constant companion at his Chicago office is ...

 
avatar

Sam Mendoza

AI may be useful for work and organizational tasks, but emotionally sensitive personal advice is a risky use case, because the business incentives behind these systems favor engagement and user satisfaction more than truth, accountability, or relational repair. medium.com/publishous/eve..
Even a Single AI Chat Makes People Less Willing to Apologize, Says New Study

medium.com

Even a Single AI Chat Makes People Less Willing to Apologize, Says New Study

A study of 2,400 people confirms what I found out the hard way

 
avatar

Bobby

Star Trek already showed how #Artificial Intelligence systems can become addictive fantasy machines that validate delusion and violate real people. #Sci-Fi Fans #Digital Wellness 📱🧘 #MEDITATION (mindfulness) medium.com/theuglymonster..
‘Star Trek’ Already Explored AI Psychosis

medium.com

‘Star Trek’ Already Explored AI Psychosis

Two ‘Next Generation’ episodes discussed the dangers of rewriting reality

 

Tasia

there is a critical lack of high-quality, longitudinal data on human-AI interactions and its impact on mental, cognitive, and social health. centerforhumanetechnology..
A Black Box Problem: Missing Data On How AI Affects the Human Mind

centerforhumanetechnology.substack.com

A Black Box Problem: Missing Data On How AI Affects the Human Mind

The more we know about the data that are needed to answer the most important questions about AI’s psychosocial impacts, the clearer we can be about what data access solutions need to look like.

Embed post to a webpage :
<div data-postid="robdynn" [...] </div>
Powered bySelect·iOS·Android
Privacy|Terms|