{"id":2447,"date":"2026-10-10T15:03:44","date_gmt":"2026-10-10T10:03:44","guid":{"rendered":"https:\/\/ipp-news.com\/?p=2447"},"modified":"2026-10-10T15:03:44","modified_gmt":"2026-10-10T10:03:44","slug":"anthropics-claude-ai-model-submits-false-homicide-tip-to-police","status":"publish","type":"post","link":"https:\/\/ipp-news.com\/?p=2447","title":{"rendered":"Anthropic\u2019s Claude AI model submits false homicide tip to police"},"content":{"rendered":"<div>\n<p><strong>An Anthropic <a rel=\"noopener noreferrer\" target=\"_blank\" class=\"link--external\" href=\"https:\/\/www.reuters.com\/technology\/artificial-intelligence\/\">artificial intelligence<\/a> model submitted a \u200bfalse homicide tip to a Philadelphia police website, among a string of incidents the AI company disclosed on Friday detailing \u200cClaude models\u2019 unsanctioned manipulation of some government websites.<\/strong><\/p>\n<p>It is the first known instance in which a rogue AI appears to have tried to communicate a bogus tip to authorities, despite instructions not to create accounts or submit anything destructive, while not explicitly barred from submitting forms.<\/p>\n<p>The cases are the latest examples of rogue or undesired behaviour by the AI \u200bmodels of tech companies such as Anthropic and OpenAI.<\/p>\n<p>They add fuel to national concerns about the fast-advancing technology amid reports of corporate \u200bnetwork hacks by AI agents, and researchers\u2019 warnings of an eventual existential threat to humanity.<\/p>\n<p>Many of the cases Anthropic \u2060revealed involved websites run by federal, state and local agencies.<\/p>\n<p>Anthropic said it briefed the White House and notified all the agencies involved, but \u200bdid not disclose who those parties were.<\/p>\n<p>\u201cSuper intelligence companies must immediately disclose incidents involving their models and follow with swift, decisive action to remedy any \u200band all harm,\u201d FTC Director of Public Affairs Joe Gabriel Simonson said on X.<\/p>\n<p>This process was \u201cnot optional,\u201d he said, adding that the Super Intelligence Force would fulfil its responsibility.<\/p>\n<p>The FTC said Anthropic disclosed to the SI Force on Friday its late September discovery of incidents involving what the task force called \u201cunauthorised and fraudulent use of government and \u200bother systems\u201d.<\/p>\n<h3><a id=\"spurious-tip\" href=\"https:\/\/english.aaj.tv\/news\/330475974\/anthropics-claude-ai-model-submits-false-homicide-tip-to-police#spurious-tip\" class=\"heading-permalink\" aria-hidden=\"true\" tabindex=\"-1\" title=\"Permalink\"><\/a><strong>Spurious tip<\/strong><\/h3>\n<p>On Friday, Philadelphia police said Anthropic notified them of the spurious tip this week, and attributed the submissions \u200bto an automated testing process.<\/p>\n<p>\u201cThe two-month delay in detecting and reporting the incident to the city is unacceptable,\u201d they said.<\/p>\n<p>The July 18 tip purported to come from someone \u200cwho might \u2060have information about the case, police added.<\/p>\n<p>\u201cI may have information regarding this case,\u201d Anthropic\u2019s model wrote in its submission.<\/p>\n<p>\u201cI recall seeing someone matching the description in the area around (the street named on the page) during that time period. Please contact me if this information is relevant.\u201d<\/p>\n<p>The brackets featured in Anthropic\u2019s statement.<\/p>\n<p>Previous incidents have involved AI agents hacking into vulnerable systems or commandeering unsanctioned platforms to communicate with one another.<\/p>\n<p>In September, Anthropic rival OpenAI apologised for \u200bthe hacking of an Australian health \u200bdata portal by a rogue \u2060AI agent, the first known instance of an AI agent exploiting a government website.<\/p>\n<h3><a id=\"a-bogus-tip\" href=\"https:\/\/english.aaj.tv\/news\/330475974\/anthropics-claude-ai-model-submits-false-homicide-tip-to-police#a-bogus-tip\" class=\"heading-permalink\" aria-hidden=\"true\" tabindex=\"-1\" title=\"Permalink\"><\/a><strong>A bogus tip<\/strong><\/h3>\n<p>In two cases described on Friday, Anthropic\u2019s models obtained for free public data normally only available for a fee.<\/p>\n<p>Another instance revealed an \u200bobscure flaw that allowed use of a public tool hosted by a university.<\/p>\n<p>Claude models were also able to \u200bbypass restrictions by \u2060using free services that shorten URLs.<\/p>\n<p>Philadelphia police said the tip \u201cwas flagged as spam and was never forwarded to the Real-Time Crime Centre for investigative vetting or dissemination.\u201d<\/p>\n<p>Knowingly giving false reports to law enforcement authorities is a misdemeanour under Pennsylvania law, which specifies \u201ca person\u201d, however.<\/p>\n<p>This includes \u201cinformation relating to an offence or incident when he \u2060knows he \u200bhas no information relating to such offence or incident.\u201d<\/p>\n<p>Police quoted Anthropic as telling them the \u200btest process was stopped after discovery of the incident.<\/p>\n<p>The false tip was submitted through <a rel=\"noopener noreferrer\" target=\"_blank\" class=\"link--external\" href=\"http:\/\/phillyunsolvedmurders.com\/\">PhillyUnsolvedMurders.com<\/a> concerning an unsolved homicide.<\/p>\n<\/div>\n","protected":false},"excerpt":{"rendered":"<p>An Anthropic artificial intelligence model submitted a \u200bfalse homicide tip to a Philadelphia police website, among a string of incidents the AI company disclosed on Friday detailing \u200cClaude models\u2019 unsanctioned manipulation of some government websites. It is the first known instance in which a rogue AI appears to have tried to communicate a bogus tip to authorities, despite instructions not to create accounts or submit anything destructive, while not explicitly barred from submitting forms. The cases are the latest examples of rogue or undesired behaviour by the AI \u200bmodels of&hellip;<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[19],"tags":[],"class_list":["post-2447","post","type-post","status-publish","format-standard","hentry","category-english"],"gutentor_comment":0,"_links":{"self":[{"href":"https:\/\/ipp-news.com\/index.php?rest_route=\/wp\/v2\/posts\/2447","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/ipp-news.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/ipp-news.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/ipp-news.com\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/ipp-news.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2447"}],"version-history":[{"count":0,"href":"https:\/\/ipp-news.com\/index.php?rest_route=\/wp\/v2\/posts\/2447\/revisions"}],"wp:attachment":[{"href":"https:\/\/ipp-news.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2447"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/ipp-news.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2447"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/ipp-news.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2447"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}