{"id":5769,"date":"2024-11-22T05:28:00","date_gmt":"2024-11-21T20:28:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=5769"},"modified":"2024-11-22T05:28:00","modified_gmt":"2024-11-21T20:28:00","slug":"jailbreak-attacks-and-defenses-against-multimodal-generative-models-a-survey","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=5769","title":{"rendered":"Jailbreak Attacks and Defenses against Multimodal Generative Models: A Survey\u00a0"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Jailbreak Attacks and Defenses against Multimodal Generative Models: A Survey\u00a0<\/strong>[50.0]<br>\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u751f\u6210\u30e2\u30c7\u30eb\u306f\u3001\u30d3\u30eb\u30c8\u30a4\u30f3\u306e\u5b89\u5168\u6a5f\u69cb\u3092\u30d0\u30a4\u30d1\u30b9\u3057\u3001\u6f5c\u5728\u7684\u306b\u6709\u5bb3\u306a\u30b3\u30f3\u30c6\u30f3\u30c4\u306e\u751f\u6210\u3092\u8a98\u5c0e\u3067\u304d\u308b\u3001\u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u653b\u6483\u306e\u5f71\u97ff\u3092\u53d7\u3051\u3084\u3059\u3044\u3002 \u672c\u8abf\u67fb\u306f,\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u751f\u6210\u30e2\u30c7\u30eb\u306b\u304a\u3051\u308b\u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u3068\u9632\u5fa1\u306b\u3064\u3044\u3066\u30ec\u30d3\u30e5\u30fc\u3059\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2411.09259v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2411.09259v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Thu, 14 Nov 2024 07:51:51 GMT)<\/li>\n\n\n\n<li>\u30de\u30eb\u30c1\u30e2\u30fc\u30c0\u30eb\u8a2d\u5b9a\u306b\u304a\u3051\u308bJailbreak\u653b\u6483\u306e\u30b5\u30fc\u30d9\u30a4\u3002\u30e2\u30c0\u30ea\u30c6\u30a3\u304c\u5897\u3048\u308b\u3068\u653b\u6483\u306b\u95a2\u3059\u308b\u30d0\u30ea\u30a8\u30fc\u30b7\u30e7\u30f3\u3082\u5897\u3048\u3001\u9762\u767d\u3044\uff08\u3068\u540c\u6642\u306b\u9632\u5fa1\u306e\u96e3\u3057\u3055\u304c\u8208\u5473\u6df1\u3044\uff09<\/li>\n\n\n\n<li>\u672c\u30b5\u30fc\u30d9\u30a4\u3067\u306f\u300c1) Input Level: Attackers and defenders operate solely on the input data.\u300d\u3001\u300c2) Encoder Level: With access to the encoder, attackers optimize adversarial inputs to inject malicious information into the encoding process, while defenders work to prevent harmful information from being encoded within the latent space\u300d\u3001\u300c3) Generator Level: With full access to the generative models, attackers leverage inference information, such as activations and gradients, and fine-tune models to increase adversarial effectiveness, \u300d\u3001\u300c4) Output Level: With the output from the generative model, attackers can iteratively refine adversarial inputs,\u300d\u3068\u3044\u3046\u30ec\u30d9\u30eb\u5206\u3051\u3092\u63a1\u7528\u3057\u3066\u3044\u308b<\/li>\n\n\n\n<li>\u30ea\u30dd\u30b8\u30c8\u30ea\u306f<a href=\"https:\/\/github.com\/liuxuannan\/Awesome-Multimodal-Jailbreak\">GitHub &#8211; liuxuannan\/Awesome-Multimodal-Jailbreak<\/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":[207,387,524],"class_list":["post-5769","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-jailbreak","tag-survey","tag-524"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5769","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=5769"}],"version-history":[{"count":0,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/5769\/revisions"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=5769"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=5769"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=5769"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}