{"id":454040,"date":"2026-02-02T18:01:11","date_gmt":"2026-02-02T18:01:11","guid":{"rendered":"https:\/\/www.newsbeep.com\/au\/454040\/"},"modified":"2026-02-02T18:01:11","modified_gmt":"2026-02-02T18:01:11","slug":"new-study-examines-how-often-ai-psychosis-actually-happens-and-the-results-are-not-good","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/au\/454040\/","title":{"rendered":"New Study Examines How Often AI Psychosis Actually Happens, and the Results Are Not Good"},"content":{"rendered":"

We\u2019ve seen plenty of evidence suggesting that prolonged use of popular AI chatbots like ChatGPT can coax some users<\/a> into spirals of paranoid and delusional behavior<\/a>.<\/p>\n

The phenomenon, dubbed \u201cAI psychosis,\u201d is a very real problem, with researchers warning of a huge wave<\/a> of severe mental health crises brought on by the tech. In extreme cases, especially involving people with pre-existing conditions<\/a>, the breaks with reality have even been linked suicides<\/a> and murder<\/a>.<\/p>\n

Now, thanks to a yet-to-be-peer-reviewed paper<\/a> published by researchers at Anthropic and the University of Toronto, we\u2019re beginning to grasp just how widespread the issue really is. <\/p>\n

The researchers set out to quantify patterns of what they called \u201cuser disempowerment\u201d in \u201creal-world [large language model] usage\u201d \u2014 including what they call \u201creality distortion,\u201d \u201cbelief distortion,\u201d and \u201caction distortion\u201d to denote a range of situations in which AI twists users\u2019 sense of reality, beliefs, or pushes them into taking actions.<\/p>\n

The results tell a damning story. The researchers found that one in 1,300 conversations out of almost 1.5 million analyzed chats with Anthropic\u2019s Claude led to reality distortion, and one in 6,000 conversations led to action distortion.<\/p>\n

To come to their conclusion, the researchers ran 1.5 million Claude conversations through an analysis tool called Clio to identify instances of \u201cdisempowerment.\u201d<\/p>\n

On the face, that may not sound like a huge proportion given the scale of the much larger dataset \u2014 but in absolute numbers, the research highlights a phenomenon that\u2019s affecting huge numbers of people.<\/p>\n

\u201cWe find the rates of severe disempowerment potential are relatively low,\u201d the researchers concluded. \u201cFor instance, severe reality distortion potential, the most common severe-level primitive, occurs in fewer than one in every thousand conversations.\u201d <\/p>\n

\u201cNevertheless, given the scale of AI usage, even these low rates translate to meaningful absolute numbers,\u201d they added. \u201cOur findings highlight the need for AI systems designed to robustly support human autonomy and flourishing.\u201d<\/p>\n

Worse yet, they found evidence that the prevalence of moderate or severe disempowerment increased between late 2024 and late 2025, indicating that the problem is growing as AI use spreads.<\/p>\n

\u201cAs exposure grows, users might become more comfortable discussing vulnerable topics or seeking advice,\u201d the researchers wrote in the blog post.<\/p>\n

Additionally, the team found that user feedback \u2014 in the form of an optional thumbs up or down button at the end of a given conversation with Claude \u2014 indicated that users \u201crate potentially disempowering interactions more favorably,\u201d according to an accompanying blog post<\/a> on Anthropic\u2019s website.<\/p>\n

In other words, users are more likely to come away satisfied when their reality or beliefs are being distorted, highlighting the role of sycophancy<\/a>, or the strong tendency of AI chatbots to validate a user\u2019s feelings and beliefs<\/a>.<\/p>\n

Plenty of fundamental questions remain. The researchers were upfront about admitting that they \u201ccan\u2019t pinpoint why\u201d the prevalence of moderate or severe disempowerment potential is growing. Their dataset is also limited to Claude consumer traffic, \u201cwhich limits generalizability.\u201d We also don\u2019t know how many of these identified cases led to real-world harm, as the research only focused on \u201cdisempowerment potential\u201d and not \u201cconfirmed harm.\u201d<\/p>\n

The team called for improved \u201cuser education\u201d to make sure people aren\u2019t giving up their full judgment to AI as \u201cmodel-side interventions are unlikely to fully address the problem.\u201d<\/p>\n

Nonetheless, the researchers say the research is only a \u201cfirst step\u201d to learn how \u201cAI might undermine human agency.\u201d<\/p>\n

\u201cWe can only address these patterns if we can measure them,\u201d they argued.<\/p>\n

More on psychosis: OnlyFans Rival Seemingly Succumbs to AI Psychosis, Which We Dare You to Try Explain to Your Parents<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"We\u2019ve seen plenty of evidence suggesting that prolonged use of popular AI chatbots like ChatGPT can coax some…\n","protected":false},"author":2,"featured_media":454041,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[256,254,255,64,63,105],"class_list":{"0":"post-454040","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-artificial-intelligence","8":"tag-ai","9":"tag-artificial-intelligence","10":"tag-artificialintelligence","11":"tag-au","12":"tag-australia","13":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/454040","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/comments?post=454040"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/454040\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media\/454041"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media?parent=454040"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/categories?post=454040"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/tags?post=454040"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}