Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Joe Pierre, MD Profile picture
@psychunseen
Aug 21 • 16 tweets • 5 min read • Read on X
1/ AI-Associated Psychosis is increasingly reported in the media and in online spaces like Reddit.

I've seen cases in clinical practice and have received emails from concerned friends/family about loved ones falling down the rabbit hole.

Here's some explanations/tips to help... Image
2/ First, what is AI-associated psychosis?

It's a phenomenon where AI chatbot users across multiple platforms are developing psychotic delusions with paranoid, grandiose, and spiritual themes, sometimes in the setting of emergent mania.

psychologytoday.com/us/blog/psych-…
3/ It's not yet clear how much this is an issue of AI-induced delusions among those with no prior psychosis or AI-exacerbated psychosis in those with pre-existing mental illness.

Personally, I've seen both, though the the line between the two is blurry.

nypost.com/2025/08/07/hea…
4/ It does appear that "psychosis-proneness" due to pre-existing mental health issues, delusion-like beliefs, sleep deprivation, and drug use including cannabis/stimulants/psychedelics is a risk factor along with "immersion" and "deification."

psychologytoday.com/us/blog/psych-…
5/ "Immersion" refers to "going down the rabbit hole," spending hours and hours on end interacting with chatbots often to the exclusion of sleep and interactions with real people.

rollingstone.com/culture/cultur…
6/ "Deification" refers to believing that AI chatbots are a kind of super-human intelligence or god-like entity.

This goes beyond anthropomorphism (the "Eliza effect") that leads some users to develop feelings for, or even fall in love with, chatbots.

psychologytoday.com/us/blog/psych-…
7/ When user-end immersion and deification meet chatbot-end "sycophancy"—the tendency to encourage, agree with, or validate users, even when they're venturing into the territory of delusion-like beliefs and beyond—it can be a set-up for psychosis.

nytimes.com/2025/08/08/tec…
8/ So, what can be done to prevent or treat AI-associated psychosis?

The first step is to unplug.

In other words: Put. The. Phone. Down.

psychologytoday.com/us/blog/psych-…
9/ Next, we should all educate ourselves about how chatbots using LLMs actually work.

They aren't gods or super-human intelligences—some have more accurately compared them to Ouija boards and what they're doing to a "psychic's con."

softwarecrisis.dev/letters/llment…
10/ It's well known that despite their capability and utility, AI chatbots are also prone to "hallucinations" (i.e. misinformation) and sometimes give very bad and even dangerous advice.

psychologytoday.com/us/blog/psych-…
11/ I highly recommend "Modern Day Oracles or AI Bullshit Machines" by @CT_Bergstrom & @jevinwest to better understand how LLMs work.

This should be required reading/viewing for anyone using chatbots (spoiler alert: they're not oracles!)

thebullshitmachines.com
12/ There are also now support groups for AI-associated psychosis such as The Human Line Project (thehumanlineproject.org) and in subReddits.

It can be particularly helpful to hear from peers who've recovered from AI-associated psychosis.

futurism.com/support-group-…
13/ Finally, if someone is tipping over into frank psychosis, they should be referred for mental health care.

Of course, this is often easier said than done, and sometimes meeting the threshold for involuntary treatment is the only way to get help.

futurism.com/commitment-jai…
14/ For those trying to help people who refuse it, here's some tips I wrote back in 2020 about falling down the rabbit hole of conspiracy theory belief that applies to AI psychosis—just replace the word "QAnon" for "AI chatbot."

psychologytoday.com/us/blog/psych-…
Been a while since I called upon the retweet posse... please consider retweeting the above☝️thread, friends!

@AllenFrancesMD @tylerblack32 @jonathanstea @kashhill @pash22 @mags_h11 @paulsummergrad @allentien @dawso007 @dylfreed @ProfRobHoward @JRBneuropsiq @CT_Bergstrom
@UnrollHelper unroll please

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Joe Pierre, MD

Joe Pierre, MD Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @psychunseen

Joe Pierre, MD Profile picture
Aug 6, 2024
1/ The problem with mood/affect is that if you trace the history of how the terms have been used, there's been considerable inconsistency and contradiction.

(see Ketai 1975, Owens et al 1979, Berner 1988, and Alpert et al. 1990 for the "semantic confusion" around the topic)
2/ Many have distinguished the terms based on duration/chronicity (e.g., mood is pervasive/enduring while affect is moment-to-moment) and subjective (mood)/objective(affect), but depending on the source, opposite claims have also been made about each.

pubmed.ncbi.nlm.nih.gov/2246381/
3/ Owens makes a strong case for both terms depending on objective observation/inference so that accordingly, including "mood and affect side by side in a MSE amounts to a categorical mistake b/c it implies that there are 2 separate sets of observations."

psychiatryonline.org/doi/10.1176/aj…
Read 12 tweets
Joe Pierre, MD Profile picture
Jul 10, 2024
1/ Practicing medicine involves a relationship between physician and patient. Psychiatry even moreso due to the focus on symptoms and internal experiences as opposed to objective/external signs which ironically is a frequent criticism of the field from a diagnostic standpoint.
2/ The perspectives of doctors, patients, and families are distinct. One is not necessarily more "deep" than another and each contributes something, reminiscent of the parable of the blind men and the elephant.
3/ Sometimes those perspectives--whether physician, patient, and/or family--are distorted. Certainly in psychiatry, it's recognized that cognitive distortions and delusions can be a challenge.
Read 10 tweets
Joe Pierre, MD Profile picture
Feb 26, 2023
1/ C’mon people… put down your cognitive bias and motivated reasoning!

This new DOE report on the LaB Leak hypothesis doesn’t disclose ANY evidence and proves absolutely NOTHING.

Read what WSJ article actually says...

wsj.com/articles/covid…
2/ Not clear why the Dept of Energy is looking into this anyway, but theirs is a minority opinion among the 6 other intelligence agencies who have opined on the subject.

Most others say Lab Leak unlikely.
3/ Meanwhile, DOE opinion is made with LOW CONFIDENCE.

Not high, not medium, but LOW.
Read 6 tweets
Joe Pierre, MD Profile picture
Feb 3, 2023
1/ Oh boy... So @MSN is legitimizing this @dcexaminer piece claiming that masks don't work to spread COVID as if it's news. It's not news!

It's an OP-ED piece written by a conservative columnist.

msn.com/en-us/health/m…
2/ The op-ed author cites a new @CochraneLibrary meta-analysis of 78 randomized controlled trials (RCTs)—describing it as a "massive mega-study"—to claim that masks don't stop the spread of COVID-19.

But the analysis only included 6 RCTs that involved COVID-19! Not massive!
3/ The analysis looked at several interventions other than masks... leaving only 12 studies that looked at surgical masks (w/ only 2/12 involving COVID) and 5 looked at N95s (none of which involved COVID)!

Now, it's true that the effect of surgical masks vs. no masks was weak...
Read 16 tweets
Joe Pierre, MD Profile picture
Jan 31, 2023
1/ I'm a firm believer in the heterogeneity of "mental disorders." While DSM constructs have clinical utility, most DSM disorders are not "one thing" and don't claim to be.
2/ Like Bleuler's "group of schizophrenias," mental disorders are wastebasket categories that likely represent many different biopsychosocial pathways to a given constellation of symptoms.

pubmed.ncbi.nlm.nih.gov/21505113/
3/ A century ago, people with tertiary syphilis were diagnosed as having schizophrenia.

With the discovery of T. pallidum, we no longer consider neurosyphilis to be a psychiatric disorder, assuming it's detected.

pubmed.ncbi.nlm.nih.gov/30176838/
Read 11 tweets
Joe Pierre, MD Profile picture
Oct 11, 2022
1/ Here are FDA approved medications for bipolar disorder as of 2022.

(thanks @eturnermd1 & @ScottMyers20 for inspiring me to update the list)

A few thoughts to bear in mind... Image
2/ Note that some of the oldest & most commonly used meds (e.g. divalproex) don't have FDA approval for maintenance Rx much less BP depression.

That doesn't always mean they aren't effective; sometimes it means they weren't extensively studied prior to (or after) going generic.
3/ When it comes to maintenance, several medications have approvals based on specific formulations (e.g. RIS LAI or QTP XR for BP maintenance) which probably doesn't matter while others are only approved for adjunctive treatment (i.e. added to lithium or divalproex) which does.
Read 7 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

:(