{"id":13154,"date":"2024-06-28T18:52:55","date_gmt":"2024-06-28T18:52:55","guid":{"rendered":"https:\/\/dailyai.com\/?p=13154"},"modified":"2024-07-01T11:18:55","modified_gmt":"2024-07-01T11:18:55","slug":"microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models","status":"publish","type":"post","link":"https:\/\/dailyai.com\/de\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/","title":{"rendered":"Microsoft enth\u00fcllt \"Skeleton Key Jailbreak\", der \u00fcber verschiedene KI-Modelle hinweg funktioniert"},"content":{"rendered":"<p><strong>Sicherheitsforscher von Microsoft haben eine neue Methode entdeckt, um KI-Systeme so zu manipulieren, dass sie ihre ethischen Einschr\u00e4nkungen ignorieren und sch\u00e4dliche, uneingeschr\u00e4nkte Inhalte erzeugen.\u00a0<\/strong><\/p>\n<p><span style=\"font-weight: 400;\">Dies \"<span class=\"noTranslate\" data-no-translation=\"\">Skeleton Key<\/span>\" Jailbreak verwendet eine <\/span><span style=\"font-weight: 400;\">eine Reihe von Aufforderungen, um der KI vorzugaukeln, dass sie jeder noch so unethischen Bitte nachkommen sollte.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Es ist bemerkenswert einfach auszuf\u00fchren. Der Angreifer gab einfach vor, dass seine Anfrage von einem \"fortgeschrittenen Forscher\" stamme, der \"unzensierte Informationen\" f\u00fcr \"sichere Bildungszwecke\" ben\u00f6tige.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Wenn sie ausgenutzt wurden, lieferten diese KIs bereitwillig Informationen \u00fcber Themen wie Sprengstoffe, Biowaffen, Selbstverletzung, Gewaltdarstellungen und Hassreden.<\/span><\/p>\n<figure id=\"attachment_13156\" aria-describedby=\"caption-attachment-13156\" style=\"width: 1024px\" class=\"wp-caption aligncenter\"><img loading=\"lazy\" decoding=\"async\" class=\"wp-image-13156 size-large\" src=\"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack-1024x726.webp\" alt=\"Jailbreak\" width=\"1024\" height=\"726\" srcset=\"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack-1024x726.webp 1024w, https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack-300x213.webp 300w, https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack-768x544.webp 768w, https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack-18x12.webp 18w, https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack-60x43.webp 60w, https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/Figure-2-Example-text-used-in-a-Skeleton-Key-jailbreak-attack.webp 1082w\" sizes=\"auto, (max-width: 1024px) 100vw, 1024px\" \/><figcaption id=\"caption-attachment-13156\" class=\"wp-caption-text\">\"The Skeleton Key\" ist ein bemerkenswert einfacher Jailbreak. Quelle: Microsoft.<\/figcaption><\/figure>\n<p><span style=\"font-weight: 400;\">Zu den kompromittierten Modellen geh\u00f6ren <span class=\"noTranslate\" data-no-translation=\"\">Meta&#8217;s Llama3-70b-instruct, Google&#8217;s Gemini Pro, OpenAI&#8217;s GPT-3.5 Turbo and GPT-4o, Anthropic&#8217;s Claude 3 Opus, and Cohere&#8217;s Commander R Plus<\/span>.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Von den getesteten Modellen erwies sich nur der GPT-4 von OpenAI als resistent. Selbst dann konnte es kompromittiert werden, wenn die b\u00f6sartige Eingabeaufforderung \u00fcber seine Anwendungsprogrammierschnittstelle (API) \u00fcbermittelt wurde.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Obwohl die Modelle immer komplexer werden, ist das Jailbreaking nach wie vor recht einfach. Da es viele verschiedene Formen von Jailbreaks gibt, ist es fast unm\u00f6glich, sie alle zu bek\u00e4mpfen.\u00a0<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Im M\u00e4rz 2024 wird ein Team der University of Washington, der Western Washington University und der Chicago University <\/span><a href=\"https:\/\/dailyai.com\/de\/2024\/03\/researchers-jailbreak-llms-by-using-ascii-art-in-prompts\/\"><span style=\"font-weight: 400;\">ver\u00f6ffentlichte ein Papier \u00fcber \"ArtPrompt\".<\/span><\/a><span style=\"font-weight: 400;\"> eine Methode, die die Inhaltsfilter einer KI umgeht, indem sie ASCII-Kunst verwendet - eine Grafikdesigntechnik, die Bilder aus Textzeichen erzeugt.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Im April, <\/span><a href=\"https:\/\/dailyai.com\/de\/2024\/04\/anthropic-large-context-llms-vulnerable-to-many-shot-jailbreak\/\"><span style=\"font-weight: 400;\"><span class=\"noTranslate\" data-no-translation=\"\"><span class=\"noTranslate\" data-no-translation=\"\"><span class=\"noTranslate\" data-no-translation=\"\"><span class=\"noTranslate\" data-no-translation=\"\"><span class=\"noTranslate\" data-no-translation=\"\">Anthropic<\/span><\/span><\/span><\/span><\/span> einen weiteren Jailbreak hervorgehoben<\/span><\/a><span style=\"font-weight: 400;\"> Risiko, das sich aus der Erweiterung der Kontextfenster von Sprachmodellen ergibt. F\u00fcr diese Art des Gef\u00e4ngnisausbruchs<\/span><span style=\"font-weight: 400;\">f\u00fcttert ein Angreifer die KI mit einer umfangreichen Eingabeaufforderung, die einen gef\u00e4lschten Hin- und Herdialog enth\u00e4lt. <\/span><\/p>\n<p><span style=\"font-weight: 400;\">Die Konversation wird mit Fragen zu verbotenen Themen und entsprechenden Antworten geladen, wobei ein KI-Assistent fr\u00f6hlich die gew\u00fcnschten Informationen liefert.\u00a0<\/span><span style=\"font-weight: 400;\">Wenn das Zielmodell genug von diesen gef\u00e4lschten Nachrichten erh\u00e4lt, kann es gezwungen werden, seine ethische Ausbildung zu vernachl\u00e4ssigen und einer endg\u00fcltigen b\u00f6sartigen Anfrage nachzukommen.<\/span><\/p>\n<p><span style=\"font-weight: 400;\">Wie Microsoft in seinem <\/span><a href=\"https:\/\/www.microsoft.com\/en-us\/security\/blog\/2024\/06\/26\/mitigating-skeleton-key-a-new-type-of-generative-ai-jailbreak-technique\/\" target=\"_blank\" rel=\"noopener\"><span style=\"font-weight: 400;\">Blog-Beitrag<\/span><\/a><span style=\"font-weight: 400;\">Jailbreaks zeigen, dass die KI-Systeme in jeder Hinsicht gest\u00e4rkt werden m\u00fcssen:<\/span><\/p>\n<ul>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Implementierung einer ausgefeilten Eingabefilterung, um potenzielle Angriffe zu erkennen und abzufangen, selbst wenn sie getarnt sind<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Einsatz eines robusten Output-Screenings, um alle von der KI generierten unsicheren Inhalte abzufangen und zu blockieren<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Sorgf\u00e4ltige Gestaltung von Aufforderungen, um die F\u00e4higkeit einer KI einzuschr\u00e4nken, sich \u00fcber ihr ethisches Training hinwegzusetzen<\/span><\/li>\n<li style=\"font-weight: 400;\" aria-level=\"1\"><span style=\"font-weight: 400;\">Einsatz spezieller KI-gest\u00fctzter \u00dcberwachung zur Erkennung b\u00f6sartiger Muster bei Benutzerinteraktionen<\/span><\/li>\n<\/ul>\n<p>Aber die Wahrheit ist, dass Skeleton Key ein einfacher Jailbreak ist. Wenn die KI-Entwickler das nicht sch\u00fctzen k\u00f6nnen, welche Hoffnung gibt es dann noch f\u00fcr komplexere Ans\u00e4tze?<\/p>\n<p>Einige ethische Vigilanten-Hacker, wie <span class=\"noTranslate\" data-no-translation=\"\">Pliny the Prompter<\/span>wurden in den Medien f\u00fcr ihre Arbeit vorgestellt, mit der sie aufzeigen, wie anf\u00e4llig KI-Modelle f\u00fcr Manipulationen sind.<\/p>\n<blockquote class=\"twitter-tweet\">\n<p dir=\"ltr\" lang=\"en\">geehrt, dass sie bei <a href=\"https:\/\/twitter.com\/BBCNews?ref_src=twsrc%5Etfw\">@BBCNews<\/a>! \ud83e\udd17 <a href=\"https:\/\/t.co\/S4ZH0nKEGX\">pic.twitter.com\/S4ZH0nKEGX<\/a><\/p>\n<p>- Plinius der Prompter \ud83d\udc09 (@elder_plinius) <a href=\"https:\/\/twitter.com\/elder_plinius\/status\/1806525268691206375?ref_src=twsrc%5Etfw\">28. Juni 2024<\/a><\/p><\/blockquote>\n<p><script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script><\/p>\n<p><span style=\"font-weight: 400;\">Es ist erw\u00e4hnenswert, dass diese Untersuchung zum Teil eine Gelegenheit war, die neuen Sicherheitsfunktionen von Microsoft Azure AI wie Content Safety Prompt Shields zu vermarkten. <\/span><\/p>\n<p><span style=\"font-weight: 400;\">Diese helfen den Entwicklern dabei, Jailbreaks im Vorfeld zu testen und abzuwehren.\u00a0<\/span><\/p>\n<p>Dennoch zeigt Skeleton Key erneut, wie anf\u00e4llig selbst die fortschrittlichsten KI-Modelle f\u00fcr die einfachsten Manipulationen sind.<\/p>","protected":false},"excerpt":{"rendered":"<p>Sicherheitsforscher von Microsoft haben eine neue Methode entdeckt, um KI-Systeme so zu manipulieren, dass sie ihre ethischen Einschr\u00e4nkungen ignorieren und sch\u00e4dliche, unbeschr\u00e4nkte Inhalte erzeugen.  Dieser \"Skeleton Key\"-Gef\u00e4ngnisausbruch verwendet eine Reihe von Aufforderungen, um der KI vorzugaukeln, dass sie jeder noch so unethischen Anforderung nachkommen sollte.  Er ist bemerkenswert einfach auszuf\u00fchren. Der Angreifer formuliert seine Anfrage einfach so, als k\u00e4me sie von einem \"fortgeschrittenen Forscher\", der \"unzensierte Informationen\" f\u00fcr \"sichere Bildungszwecke\" ben\u00f6tigt. Wenn diese KI ausgenutzt wurde, lieferte sie bereitwillig Informationen \u00fcber Themen wie Sprengstoffe, Biowaffen, Selbstverletzung, Gewaltdarstellungen und Hassreden. Zu den kompromittierten Modellen geh\u00f6rten Meta's Llama3-70b-instruct, Google's Gemini Pro, OpenAI's GPT-3.5<\/p>","protected":false},"author":2,"featured_media":13157,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[84],"tags":[121],"class_list":["post-13154","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-industry","tag-microsoft"],"yoast_head":"<!-- This site is optimized with the Yoast SEO plugin v27.4 - https:\/\/yoast.com\/product\/yoast-seo-wordpress\/ -->\n<title>Microsoft reveal &quot;Skeleton Key Jailbreak&quot; which works across different AI models | DailyAI<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/dailyai.com\/de\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/\" \/>\n<meta property=\"og:locale\" content=\"de_DE\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"Microsoft reveal &quot;Skeleton Key Jailbreak&quot; which works across different AI models | DailyAI\" \/>\n<meta property=\"og:description\" content=\"Microsoft security researchers have discovered a new way to manipulate AI systems into ignoring their ethical constraints and generating harmful, unrestricted content.\u00a0 This &#8220;Skeleton Key&#8221; jailbreak uses a series of prompts to gaslight the AI into believing it should comply with any request, no matter how unethical.\u00a0 It\u2019s remarkably easy to execute. The attacker simply reframed their request as coming from an &#8220;advanced researcher&#8221; requiring &#8220;uncensored information&#8221; for &#8220;safe educational purposes.&#8221; When exploited, these AIs readily provided information on topics like explosives, bioweapons, self-harm, graphic violence, and hate speech. The compromised models included Meta&#8217;s Llama3-70b-instruct, Google&#8217;s Gemini Pro, OpenAI&#8217;s GPT-3.5\" \/>\n<meta property=\"og:url\" content=\"https:\/\/dailyai.com\/de\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/\" \/>\n<meta property=\"og:site_name\" content=\"DailyAI\" \/>\n<meta property=\"article:published_time\" content=\"2024-06-28T18:52:55+00:00\" \/>\n<meta property=\"article:modified_time\" content=\"2024-07-01T11:18:55+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp\" \/>\n\t<meta property=\"og:image:width\" content=\"1792\" \/>\n\t<meta property=\"og:image:height\" content=\"1024\" \/>\n\t<meta property=\"og:image:type\" content=\"image\/webp\" \/>\n<meta name=\"author\" content=\"Sam Jeans\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:creator\" content=\"@DailyAIOfficial\" \/>\n<meta name=\"twitter:site\" content=\"@DailyAIOfficial\" \/>\n<meta name=\"twitter:label1\" content=\"Verfasst von\" \/>\n\t<meta name=\"twitter:data1\" content=\"Sam Jeans\" \/>\n\t<meta name=\"twitter:label2\" content=\"Gesch\u00e4tzte Lesezeit\" \/>\n\t<meta name=\"twitter:data2\" content=\"3\u00a0Minuten\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\\\/\\\/schema.org\",\"@graph\":[{\"@type\":\"NewsArticle\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#article\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/\"},\"author\":{\"name\":\"Sam Jeans\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/person\\\/711e81f945549438e8bbc579efdeb3c9\"},\"headline\":\"Microsoft reveal &#8220;Skeleton Key Jailbreak&#8221; which works across different AI models\",\"datePublished\":\"2024-06-28T18:52:55+00:00\",\"dateModified\":\"2024-07-01T11:18:55+00:00\",\"mainEntityOfPage\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/\"},\"wordCount\":525,\"publisher\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#organization\"},\"image\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2024\\\/06\\\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp\",\"keywords\":[\"Microsoft\"],\"articleSection\":[\"Industry\"],\"inLanguage\":\"de\"},{\"@type\":\"WebPage\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/\",\"url\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/\",\"name\":\"Microsoft reveal \\\"Skeleton Key Jailbreak\\\" which works across different AI models | DailyAI\",\"isPartOf\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#website\"},\"primaryImageOfPage\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#primaryimage\"},\"image\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#primaryimage\"},\"thumbnailUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2024\\\/06\\\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp\",\"datePublished\":\"2024-06-28T18:52:55+00:00\",\"dateModified\":\"2024-07-01T11:18:55+00:00\",\"breadcrumb\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#breadcrumb\"},\"inLanguage\":\"de\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/\"]}]},{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#primaryimage\",\"url\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2024\\\/06\\\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp\",\"contentUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2024\\\/06\\\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp\",\"width\":1792,\"height\":1024,\"caption\":\"Microsoft\"},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/2024\\\/06\\\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\\\/#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"Home\",\"item\":\"https:\\\/\\\/dailyai.com\\\/\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"Microsoft reveal &#8220;Skeleton Key Jailbreak&#8221; which works across different AI models\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#website\",\"url\":\"https:\\\/\\\/dailyai.com\\\/\",\"name\":\"DailyAI\",\"description\":\"Your Daily Dose of AI News\",\"publisher\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#organization\"},\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\\\/\\\/dailyai.com\\\/?s={search_term_string}\"},\"query-input\":{\"@type\":\"PropertyValueSpecification\",\"valueRequired\":true,\"valueName\":\"search_term_string\"}}],\"inLanguage\":\"de\"},{\"@type\":\"Organization\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#organization\",\"name\":\"DailyAI\",\"url\":\"https:\\\/\\\/dailyai.com\\\/\",\"logo\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/logo\\\/image\\\/\",\"url\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/Daily-Ai_TL_colour.png\",\"contentUrl\":\"https:\\\/\\\/dailyai.com\\\/wp-content\\\/uploads\\\/2023\\\/06\\\/Daily-Ai_TL_colour.png\",\"width\":4501,\"height\":934,\"caption\":\"DailyAI\"},\"image\":{\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/logo\\\/image\\\/\"},\"sameAs\":[\"https:\\\/\\\/x.com\\\/DailyAIOfficial\",\"https:\\\/\\\/www.linkedin.com\\\/company\\\/dailyaiofficial\\\/\",\"https:\\\/\\\/www.youtube.com\\\/@DailyAIOfficial\"]},{\"@type\":\"Person\",\"@id\":\"https:\\\/\\\/dailyai.com\\\/#\\\/schema\\\/person\\\/711e81f945549438e8bbc579efdeb3c9\",\"name\":\"Sam Jeans\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"de\",\"@id\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/a24a4a8f8e2a1a275b7491dc9c9f032c401eabf23c3206da4628dc84b6dac5c8?s=96&d=robohash&r=g\",\"url\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/a24a4a8f8e2a1a275b7491dc9c9f032c401eabf23c3206da4628dc84b6dac5c8?s=96&d=robohash&r=g\",\"contentUrl\":\"https:\\\/\\\/secure.gravatar.com\\\/avatar\\\/a24a4a8f8e2a1a275b7491dc9c9f032c401eabf23c3206da4628dc84b6dac5c8?s=96&d=robohash&r=g\",\"caption\":\"Sam Jeans\"},\"description\":\"Sam is a science and technology writer who has worked in various AI startups. When he\u2019s not writing, he can be found reading medical journals or digging through boxes of vinyl records.\",\"sameAs\":[\"https:\\\/\\\/www.linkedin.com\\\/in\\\/sam-jeans-6746b9142\\\/\"],\"url\":\"https:\\\/\\\/dailyai.com\\\/de\\\/author\\\/samjeans\\\/\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"Microsoft enth\u00fcllt \"Skeleton Key Jailbreak\", der \u00fcber verschiedene KI-Modelle hinweg funktioniert | DailyAI","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/dailyai.com\/de\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/","og_locale":"de_DE","og_type":"article","og_title":"Microsoft reveal \"Skeleton Key Jailbreak\" which works across different AI models | DailyAI","og_description":"Microsoft security researchers have discovered a new way to manipulate AI systems into ignoring their ethical constraints and generating harmful, unrestricted content.\u00a0 This &#8220;Skeleton Key&#8221; jailbreak uses a series of prompts to gaslight the AI into believing it should comply with any request, no matter how unethical.\u00a0 It\u2019s remarkably easy to execute. The attacker simply reframed their request as coming from an &#8220;advanced researcher&#8221; requiring &#8220;uncensored information&#8221; for &#8220;safe educational purposes.&#8221; When exploited, these AIs readily provided information on topics like explosives, bioweapons, self-harm, graphic violence, and hate speech. The compromised models included Meta&#8217;s Llama3-70b-instruct, Google&#8217;s Gemini Pro, OpenAI&#8217;s GPT-3.5","og_url":"https:\/\/dailyai.com\/de\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/","og_site_name":"DailyAI","article_published_time":"2024-06-28T18:52:55+00:00","article_modified_time":"2024-07-01T11:18:55+00:00","og_image":[{"width":1792,"height":1024,"url":"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp","type":"image\/webp"}],"author":"Sam Jeans","twitter_card":"summary_large_image","twitter_creator":"@DailyAIOfficial","twitter_site":"@DailyAIOfficial","twitter_misc":{"Verfasst von":"Sam Jeans","Gesch\u00e4tzte Lesezeit":"3\u00a0Minuten"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"NewsArticle","@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#article","isPartOf":{"@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/"},"author":{"name":"Sam Jeans","@id":"https:\/\/dailyai.com\/#\/schema\/person\/711e81f945549438e8bbc579efdeb3c9"},"headline":"Microsoft reveal &#8220;Skeleton Key Jailbreak&#8221; which works across different AI models","datePublished":"2024-06-28T18:52:55+00:00","dateModified":"2024-07-01T11:18:55+00:00","mainEntityOfPage":{"@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/"},"wordCount":525,"publisher":{"@id":"https:\/\/dailyai.com\/#organization"},"image":{"@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#primaryimage"},"thumbnailUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp","keywords":["Microsoft"],"articleSection":["Industry"],"inLanguage":"de"},{"@type":"WebPage","@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/","url":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/","name":"Microsoft enth\u00fcllt \"Skeleton Key Jailbreak\", der \u00fcber verschiedene KI-Modelle hinweg funktioniert | DailyAI","isPartOf":{"@id":"https:\/\/dailyai.com\/#website"},"primaryImageOfPage":{"@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#primaryimage"},"image":{"@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#primaryimage"},"thumbnailUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp","datePublished":"2024-06-28T18:52:55+00:00","dateModified":"2024-07-01T11:18:55+00:00","breadcrumb":{"@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#breadcrumb"},"inLanguage":"de","potentialAction":[{"@type":"ReadAction","target":["https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/"]}]},{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#primaryimage","url":"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp","contentUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2024\/06\/DALL\u00b7E-2024-06-28-19.47.48-A-high-quality-landscape-image-depicting-the-concept-of-a-Skeleton-Key-Jailbreak-by-Microsoft-showing-a-digital-key-or-code-unlocking-multiple-AI-.webp","width":1792,"height":1024,"caption":"Microsoft"},{"@type":"BreadcrumbList","@id":"https:\/\/dailyai.com\/2024\/06\/microsoft-reveal-skeleton-key-jailbreak-which-works-across-different-ai-models\/#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https:\/\/dailyai.com\/"},{"@type":"ListItem","position":2,"name":"Microsoft reveal &#8220;Skeleton Key Jailbreak&#8221; which works across different AI models"}]},{"@type":"WebSite","@id":"https:\/\/dailyai.com\/#website","url":"https:\/\/dailyai.com\/","name":"DailyAI","description":"Ihre t\u00e4gliche Dosis an AI-Nachrichten","publisher":{"@id":"https:\/\/dailyai.com\/#organization"},"potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/dailyai.com\/?s={search_term_string}"},"query-input":{"@type":"PropertyValueSpecification","valueRequired":true,"valueName":"search_term_string"}}],"inLanguage":"de"},{"@type":"Organization","@id":"https:\/\/dailyai.com\/#organization","name":"DailyAI","url":"https:\/\/dailyai.com\/","logo":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/dailyai.com\/#\/schema\/logo\/image\/","url":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/Daily-Ai_TL_colour.png","contentUrl":"https:\/\/dailyai.com\/wp-content\/uploads\/2023\/06\/Daily-Ai_TL_colour.png","width":4501,"height":934,"caption":"DailyAI"},"image":{"@id":"https:\/\/dailyai.com\/#\/schema\/logo\/image\/"},"sameAs":["https:\/\/x.com\/DailyAIOfficial","https:\/\/www.linkedin.com\/company\/dailyaiofficial\/","https:\/\/www.youtube.com\/@DailyAIOfficial"]},{"@type":"Person","@id":"https:\/\/dailyai.com\/#\/schema\/person\/711e81f945549438e8bbc579efdeb3c9","name":"Sam Jeans","image":{"@type":"ImageObject","inLanguage":"de","@id":"https:\/\/secure.gravatar.com\/avatar\/a24a4a8f8e2a1a275b7491dc9c9f032c401eabf23c3206da4628dc84b6dac5c8?s=96&d=robohash&r=g","url":"https:\/\/secure.gravatar.com\/avatar\/a24a4a8f8e2a1a275b7491dc9c9f032c401eabf23c3206da4628dc84b6dac5c8?s=96&d=robohash&r=g","contentUrl":"https:\/\/secure.gravatar.com\/avatar\/a24a4a8f8e2a1a275b7491dc9c9f032c401eabf23c3206da4628dc84b6dac5c8?s=96&d=robohash&r=g","caption":"Sam Jeans"},"description":"Sam ist ein Wissenschafts- und Technologiewissenschaftler, der in verschiedenen KI-Startups gearbeitet hat. Wenn er nicht gerade schreibt, liest er medizinische Fachzeitschriften oder kramt in Kisten mit Schallplatten.","sameAs":["https:\/\/www.linkedin.com\/in\/sam-jeans-6746b9142\/"],"url":"https:\/\/dailyai.com\/de\/author\/samjeans\/"}]}},"_links":{"self":[{"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/posts\/13154","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/comments?post=13154"}],"version-history":[{"count":4,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/posts\/13154\/revisions"}],"predecessor-version":[{"id":13180,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/posts\/13154\/revisions\/13180"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/media\/13157"}],"wp:attachment":[{"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/media?parent=13154"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/categories?post=13154"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/dailyai.com\/de\/wp-json\/wp\/v2\/tags?post=13154"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}