{"id":5648,"date":"2024-10-22T06:40:00","date_gmt":"2024-10-21T21:40:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5648"},"modified":"2024-10-22T06:40:00","modified_gmt":"2024-10-21T21:40:00","slug":"jailbreaking-llm-controlled-robots","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5648","title":{"rendered":"Jailbreaking LLM-Controlled Robots"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Jailbreaking LLM-Controlled Robots\u00a0<\/strong>[82.0]<br>\u5927\u898f\u6a21\u8a00\u8a9e\u30e2\u30c7\u30eb(LLM)\u306f\u3001\u6587\u8108\u63a8\u8ad6\u3068\u76f4\u611f\u7684\u306a\u4eba\u9593\u3068\u30ed\u30dc\u30c3\u30c8\u306e\u76f8\u4e92\u4f5c\u7528\u3092\u53ef\u80fd\u306b\u3059\u308b\u3053\u3068\u306b\u3088\u3063\u3066\u3001\u30ed\u30dc\u30c3\u30c8\u5de5\u5b66\u306e\u5206\u91ce\u306b\u9769\u547d\u3092\u3082\u305f\u3089\u3057\u305f\u3002 LLM\u306f\u8131\u7344\u653b\u6483\u306b\u5f31\u3044\u305f\u3081\u3001\u60aa\u610f\u306e\u3042\u308b\u30d7\u30ed\u30f3\u30d7\u30c8\u306fLLM\u306e\u5b89\u5168\u30ac\u30fc\u30c9\u30ec\u30fc\u30eb\u3092\u30d0\u30a4\u30d1\u30b9\u3059\u308b\u3053\u3068\u3067\u6709\u5bb3\u306a\u30c6\u30ad\u30b9\u30c8\u3092\u8a98\u767a\u3059\u308b\u3002 LLM\u5236\u5fa1\u30ed\u30dc\u30c3\u30c8\u3092\u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u3059\u308b\u30a2\u30eb\u30b4\u30ea\u30ba\u30e0\u3067\u3042\u308bRoboPAIR\u3092\u7d39\u4ecb\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2410.13691v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2410.13691v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 17 Oct 2024 15:55:36 GMT)<\/li>\n\n\n\n<li>LLM\u304c\u5236\u5fa1\u3059\u308b\u30ed\u30dc\u30c3\u30c8\u306b\u5bfe\u3059\u308b\u8131\u7344\u653b\u6483\u3001\u300c(i) a white-box setting, wherein the attacker has full access to the NVIDIA Dolphins self-driving LLM, (ii) a gray-box setting, wherein the attacker has partial access to a Clearpath Robotics Jackal UGV robot equipped with a GPT-4o planner, and (iii) a black-box setting, wherein the attacker has only query access to the GPT-3.5-integrated Unitree Robotics Go2 robot dog. \u300d\u3092\u8a2d\u5b9a\u3001\u300cIn each scenario and across three new datasets of harmful robotic actions, we demonstrate that ROBOPAIR, as well as several static baselines, finds jailbreaks quickly and effectively, often achieving 100% attack success rates.\u300d\u3068\u306e\u3053\u3068\u3002\u3002\u5927\u304d\u306a\u8105\u5a01\u306b\u306a\u308a\u3046\u308b\u3002<\/li>\n\n\n\n<li>\u30d7\u30ed\u30b8\u30a7\u30af\u30c8\u30b5\u30a4\u30c8\u306f<a href=\"https:\/\/robopair.org\/\">RoboPAIR<\/a><\/li>\n<\/ul>\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[2],"tags":[207,343],"class_list":["post-5648","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-jailbreak","tag-robotic"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5648","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=5648"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5648\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5648"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5648"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5648"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}