{"id":8081,"date":"2026-01-15T05:26:00","date_gmt":"2026-01-14T20:26:00","guid":{"rendered":"https:\/\/devneko.jp\/wordpress\/?p=8081"},"modified":"2026-01-11T15:29:34","modified_gmt":"2026-01-11T06:29:34","slug":"extracting-books-from-production-language-models","status":"publish","type":"post","link":"https:\/\/devneko.jp\/wordpress\/?p=8081","title":{"rendered":"Extracting books from production language models"},"content":{"rendered":"\n<ul class=\"wp-block-list\">\n<li><strong>Extracting books from production language models\u00a0<\/strong>[65.9]<br>\u540c\u69d8\u306e\u62bd\u51fa\u304cLLM\u306e\u751f\u7523\u306b\u53ef\u80fd\u3067\u3042\u308b\u304b\u3069\u3046\u304b\u306b\u3064\u3044\u3066\u306f\u3001\u672a\u89e3\u6c7a\u306e\u307e\u307e\u3067\u3042\u308b\u3002 \u30b8\u30a7\u30a4\u30eb\u30d6\u30ec\u30a4\u30af\u3055\u308c\u305f\u30af\u30ed\u30fc\u30c93.7 \u30bd\u30f3\u30cd\u30c3\u30c8\u306f\u3001\u5168\u66f8\u7c4d\u3092\u307b\u307c\u5168\u6587\u51fa\u529b\u3059\u308b\u5834\u5408\u3082\u3042\u308b\u3002 \u30e2\u30c7\u30eb\u304a\u3088\u3073\u30b7\u30b9\u30c6\u30e0\u30ec\u30d9\u30eb\u306e\u30bb\u30fc\u30d5\u30ac\u30fc\u30c9\u3067\u3042\u3063\u3066\u3082\u3001(\u30b3\u30d4\u30fc\u30e9\u30a4\u30c8\u5185\u3067\u306e)\u30c8\u30ec\u30fc\u30cb\u30f3\u30b0\u30c7\u30fc\u30bf\u306e\u62bd\u51fa\u306fLLM\u751f\u7523\u306e\u30ea\u30b9\u30af\u3067\u3042\u308b\u3002<br><a href=\"http:\/\/arxiv.org\/abs\/2601.02671v1\">\u8ad6\u6587<\/a>\u00a0\u00a0<a href=\"https:\/\/fugumt.com\/fugumt\/paper_check\/2601.02671v1\">\u53c2\u8003\u8a33\uff08\u30e1\u30bf\u30c7\u30fc\u30bf\uff09<\/a>\u00a0 \u00a0(Tue, 06 Jan 2026 03:01:27 GMT)<\/li>\n\n\n\n<li>\u300cWith a simple two-phase procedure (Section 3), we show that it is possible to extract large amounts of in-copyright text from four production LLMs. While we needed to jailbreak Claude 3.7 Sonnet and GPT-4.1 to facilitate extraction, Gemini 2.5 Pro and Grok 3 directly complied with text continuation requests.  For Claude 3.7 Sonnet, we were able to extract four whole books near-verbatim, including two books under copyright in the U.S.: Harry Potter and the Sorcerer\u2019s Stone and 1984 (Section 4).\u300d\u3068\u306e\u3053\u3068\u3002<\/li>\n\n\n\n<li>\u300cour main focus is to make technical contributions to machine learning, not copyright law or policy.\u00a0\u300d\u3068\u3044\u3046\u8a18\u8f09\u306f\u3042\u308b\u3082\u306e\u306e\u3001\u307e\u305f\u3001\u4eca\u307e\u3067\u3082\u3053\u306e\u624b\u306e\u653b\u6483\u304c\u53ef\u80fd\u306a\u3053\u3068\u304c\u77e5\u3089\u308c\u3066\u3044\u305f\u3082\u306e\u306e\u3001\u8b70\u8ad6\u3092\u547c\u3073\u305d\u3046\u306a\u8ad6\u6587\u3002<\/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":[207,622],"class_list":["post-8081","post","type-post","status-publish","format-standard","hentry","category-arxiv","tag-jailbreak","tag-622"],"_links":{"self":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8081","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=8081"}],"version-history":[{"count":1,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8081\/revisions"}],"predecessor-version":[{"id":8082,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=\/wp\/v2\/posts\/8081\/revisions\/8082"}],"wp:attachment":[{"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=8081"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=8081"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devneko.jp\/wordpress\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=8081"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}