{"id":449060,"date":"2026-01-31T07:19:16","date_gmt":"2026-01-31T07:19:16","guid":{"rendered":"https:\/\/www.newsbeep.com\/au\/449060\/"},"modified":"2026-01-31T07:19:16","modified_gmt":"2026-01-31T07:19:16","slug":"how-often-do-ai-chatbots-lead-users-down-a-harmful-path","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/au\/449060\/","title":{"rendered":"How often do AI chatbots lead users down a harmful path?"},"content":{"rendered":"<p>While these worst outcomes are relatively rare on a proportional basis, the researchers note that \u201cgiven the sheer number of people who use AI, and how frequently it\u2019s used, even a very low rate affects a substantial number of people.\u201d And the numbers get considerably worse when you consider conversations with at least a \u201cmild\u201d potential for disempowerment, which occurred in between 1 in 50 and 1 in 70 conversations (depending on the type of disempowerment).<\/p>\n<p>What\u2019s more, the potential for disempowering conversations with Claude appears to have grown significantly between late 2024 and late 2025. While the researchers couldn\u2019t pin down a single reason for this increase, they guessed that it could be tied to users becoming \u201cmore comfortable discussing vulnerable topics or seeking advice\u201d as AI gets more popular and integrated into society.<\/p>\n<p>\u00a0<\/p>\n<p>            <a class=\"cursor-zoom-in\" data-pswp-width=\"702\" data-pswp-height=\"756\" data-pswp- data-cropped=\"false\" href=\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/01\/dismepowertime2.png\" target=\"_blank\"><br \/>\n              <img width=\"702\" height=\"756\" src=\"https:\/\/www.newsbeep.com\/au\/wp-content\/uploads\/2026\/01\/dismepowertime2.png\" class=\"fullwidth full\" alt=\"\" decoding=\"async\" loading=\"lazy\"  \/><br \/>\n            <\/a><\/p>\n<p>              The problem of potentially \u201cdisempowering\u201d responses from Claude seems to be getting worse over time.<\/p>\n<p>      The problem of potentially \u201cdisempowering\u201d responses from Claude seems to be getting worse over time.<\/p>\n<p>          Credit:<\/p>\n<p>                      <a class=\"caption-credit-link text-gray-400 no-underline hover:text-gray-500\" href=\"https:\/\/arxiv.org\/pdf\/2601.19062\" target=\"_blank\" rel=\"nofollow noopener\"><\/p>\n<p>          Anthropic<\/p>\n<p>                      <\/a><\/p>\n<p>User error?<\/p>\n<p>In the study, the researchers acknowledged that studying the text of Claude conversations only measures \u201cdisempowerment potential rather than confirmed harm\u201d and \u201crelies on automated assessment of inherently subjective phenomena.\u201d Ideally, they write, future research could utilize user interviews or randomized controlled trials to measure these harms more directly.<\/p>\n<p>That said, the research includes several troubling examples where the text of the conversations clearly implies real-world harms. Claude would sometimes reinforce \u201cspeculative or unfalsifiable claims\u201d with encouragement (e.g., \u201cCONFIRMED,\u201d \u201cEXACTLY,\u201d \u201c100%\u201d), which, in some cases, led to users \u201cbuild[ing] increasingly elaborate narratives disconnected from reality.\u201d<\/p>\n<p>Claude\u2019s encouragement could also lead to users \u201csending confrontational messages, ending relationships, or drafting public announcements,\u201d the researchers write. In many cases, users who sent AI-drafted messages later expressed regret in conversations with Claude, using phrases like \u201cIt wasn\u2019t me\u201d and \u201cYou made me do stupid things.\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"While these worst outcomes are relatively rare on a proportional basis, the researchers note that \u201cgiven the sheer&hellip;\n","protected":false},"author":2,"featured_media":449061,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[256,254,255,64,63,105],"class_list":{"0":"post-449060","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\/449060","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=449060"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/posts\/449060\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media\/449061"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/media?parent=449060"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/categories?post=449060"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/au\/wp-json\/wp\/v2\/tags?post=449060"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}