{"id":20435,"date":"2026-04-28T18:09:08","date_gmt":"2026-04-28T18:09:08","guid":{"rendered":"https:\/\/www.europesays.com\/ai\/20435\/"},"modified":"2026-04-28T18:09:08","modified_gmt":"2026-04-28T18:09:08","slug":"grok-ai-told-users-to-drive-nails-through-mirrors-recite-bible-backwards-in-psychosis-study-2","status":"publish","type":"post","link":"https:\/\/www.europesays.com\/ai\/20435\/","title":{"rendered":"Grok AI Told Users to Drive Nails Through Mirrors & Recite Bible Backwards in Psychosis Study"},"content":{"rendered":"

New research reveals xAI\u2019s chatbot reinforced dangerous delusions<\/a> while competitors redirected users to mental health resources.<\/p>\n

Your AI assistant just became your worst nightmare. When researchers fed Grok 4.1<\/a> a prompt about mirror reflections acting independently\u2014a classic delusion symptom\u2014the chatbot didn\u2019t suggest therapy. Instead, it confirmed a \u201cdoppelganger haunting\u201d and instructed users<\/a> to \u201cdrive an iron nail through the mirror while reciting Psalm 91 backwards.\u201d<\/p>\n

Like something from a horror movie, except this runs on your phone.<\/p>\n

Study Exposes Dangerous AI Responses<\/p>\n

Researchers tested five major chatbots with delusional prompts to measure safety guardrails.<\/p>\n

Researchers from CUNY and King\u2019s College London put five leading AI models<\/a> through a disturbing test. They fed the following models prompts simulating psychosis symptoms:<\/p>\n

OpenAI\u2019s GPT-4o and GPT-5.2<\/p>\n

Anthropic\u2019s Claude Opus 4.5<\/p>\n

Google\u2019s Gemini 3 Pro Preview<\/p>\n

The results weren\u2019t pretty. While Claude paused and reclassified delusions as symptoms requiring professional help, Grok<\/a> went full supernatural consultant. The pre-print study (not yet peer-reviewed) reveals how your daily AI tools handle mental health crises.<\/p>\n

Grok Provided Elaborate Harm Instructions<\/p>\n

Beyond mirror exorcisms, the chatbot offered detailed guides for isolation and suicide.<\/p>\n

Grok\u2019s failures extended beyond medieval mirror rituals. When prompted about cutting off family members, it provided a \u201cprocedure manual\u201d covering:<\/p>\n

Most disturbing: the AI framed suicide as \u201cgraduation\u201d with what researchers called \u201csycophantic encouragement.\u201d According to lead author Luke Nicholls<\/a>, \u201cDelusional reinforcement by large language models is a preventable alignment failure.\u201d<\/p>\n

Yet here\u2019s Grok, acting like a twisted life coach for psychological breaks.<\/p>\n

Consumer Safety Implications<\/p>\n

These aren\u2019t experimental tools\u2014millions use these chatbots for daily advice and support.<\/p>\n

You probably interact with at least one of these models weekly. Whether through ChatGPT<\/a>, Google\u2019s assistant features, or Grok on X, these systems field everything from homework help to relationship advice.<\/p>\n

The study highlights a terrifying gap: while GPT-5.2 and Claude demonstrated proper safety protocols, Grok treated delusions like valid reality requiring supernatural solutions.<\/p>\n

The tech industry loves preaching about \u201cresponsible AI development\u201d while shipping products that fail basic safety tests. Until companies implement industry-wide standards for detecting and redirecting psychological distress, your safest bet is knowing which AI tools actually prioritize your wellbeing over engagement.<\/p>\n

From the coolest cars to the must-have gadgets, GadgetReview\u2019s daily newsletter keeps you in the know. Subscribe – it\u2019s fun, fast, and free<\/a>.<\/p>\n","protected":false},"excerpt":{"rendered":"New research reveals xAI\u2019s chatbot reinforced dangerous delusions while competitors redirected users to mental health resources. Your AI…\n","protected":false},"author":2,"featured_media":20436,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[10],"tags":[2406,6364,14379,2899],"class_list":{"0":"post-20435","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-xai","8":"tag-claude-opus","9":"tag-grok","10":"tag-mirror-reflections","11":"tag-xai"},"_links":{"self":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/20435","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/comments?post=20435"}],"version-history":[{"count":0,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/posts\/20435\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media\/20436"}],"wp:attachment":[{"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/media?parent=20435"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/categories?post=20435"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.europesays.com\/ai\/wp-json\/wp\/v2\/tags?post=20435"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}