{"id":3714,"date":"2023-08-23T05:54:00","date_gmt":"2023-08-22T20:54:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=3714"},"modified":"2023-08-23T05:54:00","modified_gmt":"2023-08-22T20:54:00","slug":"gpt-4-is-too-smart-to-be-safe-stealthy-chat-with-llms-via-cipher","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=3714","title":{"rendered":"GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher\u00a0<\/strong>[89.5]<br>\u5b9f\u9a13\u306b\u3088\u308a\u3001\u3044\u304f\u3064\u304b\u306e\u5b89\u5168\u9818\u57df\u306b\u304a\u3044\u3066\u3001GPT-4\u306e\u5b89\u5168\u6027\u30a2\u30e9\u30a4\u30e1\u30f3\u30c8\u3092\u30d0\u30a4\u30d1\u30b9\u3059\u308b\u305f\u3081\u306b\u3001\u3042\u308b\u6697\u53f7\u304c\u307b\u307c100%\u306e\u6642\u9593\u3067\u6210\u529f\u3059\u308b\u3053\u3068\u304c\u793a\u3055\u308c\u305f\u3002 \u672c\u7a3f\u3067\u306f,\u30ed\u30fc\u30eb\u30d7\u30ec\u30a4\u306e\u307f\u3092\u4f7f\u7528\u3057,\u81ea\u7136\u8a00\u8a9e\u306b\u3088\u308b\u3044\u304f\u3064\u304b\u306e\u5b9f\u6f14\u3092\u884c\u3044,\u3053\u308c\u3092\u8a98\u767a\u3059\u308b\u65b0\u3057\u3044SelfCipher\u3092\u63d0\u6848\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2308.06463v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2308.06463v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Sat, 12 Aug 2023 04:05:57 GMT)<\/li>\n\n\n\n<li>\u6697\u53f7\u3092\u4ecb\u3059\u308b\u4e8b\u3067GPT-4\u306e\u5b89\u5168\u5bfe\u7b56\u3092\u56de\u907f\u3067\u304d\u305f\u3068\u3044\u3046\u5831\u544a\u3002\u6700\u521d\u306bLLM\u306b\u6697\u53f7\u5316\u3068\u5fa9\u53f7\u306e\u30eb\u30fc\u30eb\u3092\u6559\u3048\u3053\u3080\u30d7\u30ed\u30bb\u30b9\u306a\u306e\u304c\u9762\u767d\u3044\u3002<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/RobustNLP\/CipherChat\">GitHub &#8211; RobustNLP\/CipherChat: A framework to evaluate the generalization capability of safety alignment for LLMs<\/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":[32,175],"class_list":["post-3714","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-attack","tag-gpt-4"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3714","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=3714"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/3714\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=3714"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=3714"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=3714"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}