{ "schemaVersion": "2.0", "generatedAt": "2026-09-12T13:44:57.241Z", "compatibility": { "legacyMatterFieldsRetained": true, "issueDocumentsAreNotTimelineEvents": true }, "matter": { "id": "matter-06", "slug": "thomson-reuters-v-ross-intelligence", "titleZh": "汤森路透诉 ROSS Intelligence 案", "formalName": "Thomson Reuters Enterprise Centre GmbH et al. v. ROSS Intelligence Inc.", "aliases": [ "Westlaw v. ROSS", "Thomson Reuters v. ROSS" ], "summary": "汤森路透主张 ROSS 为训练法律检索系统使用了由 Westlaw 内容制作的问答材料。地区法院在 2025 年就部分作品判定侵权并否定合理使用抗辩。", "professionalSummary": "裁判围绕 Westlaw 头注的原创性、实质性相似与四项合理使用因素展开；其事实结构与通用生成式模型训练并不完全相同。", "whyItMatters": "这是美国法院在 AI 训练语境中较早判定权利人胜诉的实体裁判。", "status": "on-appeal", "primaryTopic": "copyright", "secondaryTopics": [ "competition" ], "aiRoles": [ "training-data" ], "jurisdiction": "美国·联邦·特拉华区", "industries": [ "法律科技", "专业出版" ], "significance": { "legalNovelty": 29, "impactScope": 23, "institutionalLevel": 18, "crossBorder": 5, "sustainedCoverage": 15, "total": 90 }, "highProfile": true, "featured": true, "firstPublishedAt": "2026-07-15", "lastUpdatedAt": "2026-07-16", "proceedings": [ { "id": "matter-06:proceeding-1", "matterId": "matter-06", "formalName": "Thomson Reuters Enterprise Centre GmbH et al. v. ROSS Intelligence Inc.", "identifier": "1:20-cv-00613-SB", "identifiers": { "docketNumber": "1:20-cv-00613-SB" }, "forum": "U.S. District Court for the District of Delaware", "jurisdiction": "美国·联邦·特拉华区", "type": "civil-litigation", "stage": "judgment", "status": "active", "startedAt": "2025-02-11", "precedentialWeight": "pending-appeal", "officialUrl": "https://www.govinfo.gov/app/details/USCOURTS-ded-1_20-cv-00613/USCOURTS-ded-1_20-cv-00613-5" }, { "id": "matter-06:proceeding-2", "matterId": "matter-06", "formalName": "Thomson Reuters Enterprise Centre GmbH et al. v. ROSS Intelligence Inc. — interlocutory appeal", "identifier": "25-2153", "identifiers": { "docketNumber": "25-2153" }, "forum": "U.S. Court of Appeals for the Third Circuit", "jurisdiction": "美国·联邦·特拉华区", "type": "civil-litigation", "stage": "appeal", "status": "on-appeal", "startedAt": "2025-06-24", "precedentialWeight": "pending-appeal", "officialUrl": "https://dockets.justia.com/docket/circuit-courts/ca3/25-2153" } ], "events": [ { "id": "matter-06:event-1", "matterId": "matter-06", "proceedingId": "matter-06:proceeding-1", "date": "2025-02-11", "type": "judgment", "title": "地区法院作出修订后的简易判决意见", "summary": "地区法院就选定 Westlaw 头注认定 ROSS 构成侵权，并在该记录上否定合理使用。", "documentIds": [ "matter-06:document-1" ], "claimIds": [ "matter-06:claim-1", "matter-06:claim-procedure", "matter-06:claim-event" ] } ], "documents": [ { "id": "matter-06:document-1", "familyId": "matter-06:family-1", "matterId": "matter-06", "eventId": "matter-06:event-1", "title": "Thomson Reuters v. ROSS Intelligence — Memorandum Opinion", "originalTitle": "Thomson Reuters v. ROSS Intelligence — Memorandum Opinion", "url": "https://www.govinfo.gov/app/details/USCOURTS-ded-1_20-cv-00613/USCOURTS-ded-1_20-cv-00613-5", "kind": "official-document", "sourceId": "govinfo", "publishedAt": "2025-02-11", "primaryOrSecondary": "primary", "relationship": "independent", "identifiers": { "docketNumber": "1:20-cv-00613-SB" }, "language": "en" } ], "claims": [ { "id": "matter-06:claim-1", "matterId": "matter-06", "eventId": "matter-06:event-1", "text": "地区法院就选定 Westlaw 头注认定 ROSS 构成侵权，并在该记录上否定合理使用。", "kind": "legal-holding", "status": "confirmed", "documentIds": [ "matter-06:document-1" ], "asOf": "2025-02-11" }, { "id": "matter-06:claim-procedure", "matterId": "matter-06", "eventId": "matter-06:event-1", "text": "1:20-cv-00613-SB 是本站记录的正式程序编号，程序由 U.S. District Court for the District of Delaware 处理。", "kind": "procedural-fact", "status": "confirmed", "documentIds": [ "matter-06:document-1" ], "asOf": "2025-02-11" }, { "id": "matter-06:claim-event", "matterId": "matter-06", "eventId": "matter-06:event-1", "text": "2025-02-11，案卷记录“地区法院作出修订后的简易判决意见”这一程序节点；该日期与事件由所引直接材料支持。", "kind": "procedural-fact", "status": "confirmed", "documentIds": [ "matter-06:document-1" ], "asOf": "2025-02-11" } ], "caseBackground": { "narrativeZh": "汤森路透指控 ROSS Intelligence 通过由 Westlaw 内容制作的问答材料训练法律检索系统，侵犯 Westlaw 头注版权。特拉华地区法院在 2025 年就选定头注认定侵权并否定合理使用，裁判进入第三巡回法院中间上诉。", "partiesZh": [ "Thomson Reuters 相关实体", "ROSS Intelligence Inc." ], "aiSystemZh": "ROSS 法律检索系统", "challengedConductZh": "使用基于 Westlaw 头注制作的问答材料训练竞争性法律检索系统。", "allegedHarmZh": "权利人主张受保护头注被复制并用于建立竞争产品，损害内容与检索服务市场。", "proceduralOriginZh": "案件以 1:20-cv-00613-SB 在特拉华区提起；2025 年 2 月 11 日作出修订后的部分简易判决意见，后以 25-2153 进入第三巡回法院。", "claimIds": [ "matter-06:claim-1", "matter-06:claim-procedure", "matter-06:claim-event" ], "documentIds": [ "matter-06:document-1" ] }, "issues": [ { "id": "matter-06:issue-headnote-originality", "matterId": "matter-06", "titleZh": "Westlaw 头注的原创性", "questionZh": "被选中的 Westlaw 头注是否包含足以受保护的原创表达？", "contextZh": "头注基于司法意见制作，但权利主张针对编辑人员形成的表达，而不是法院意见本身。", "topic": "copyright", "status": "partly-resolved", "courtTreatmentZh": "地区法院在选定头注范围内认可可保护性并据此认定侵权；该结论处于中间上诉状态。", "claimIds": [ "matter-06:claim-1" ], "positions": [ { "id": "matter-06:issue-headnote-originality:position-1", "issueId": "matter-06:issue-headnote-originality", "sideZh": "汤森路透", "positionZh": "主张编辑制作的选定头注包含原创表达。", "claimIds": [], "documentIds": [] }, { "id": "matter-06:issue-headnote-originality:position-2", "issueId": "matter-06:issue-headnote-originality", "sideZh": "ROSS Intelligence", "positionZh": "在地区法院未能阻止选定头注被认定受保护，相关问题现处于上诉程序。", "claimIds": [], "documentIds": [] } ] }, { "id": "matter-06:issue-copying-similarity", "matterId": "matter-06", "titleZh": "训练材料与头注的实质性相似", "questionZh": "ROSS 使用的问答训练材料与选定 Westlaw 头注之间是否存在可归责的复制和实质性相似？", "contextZh": "训练材料并非直接标为 Westlaw 数据，法院仍需把具体问答与具体头注相互对应。", "topic": "copyright", "status": "partly-resolved", "courtTreatmentZh": "地区法院在选定作品和当时记录上认定复制及侵权成立，未把结论扩展到所有 Westlaw 内容。", "claimIds": [ "matter-06:claim-1" ], "positions": [ { "id": "matter-06:issue-copying-similarity:position-1", "issueId": "matter-06:issue-copying-similarity", "sideZh": "汤森路透", "positionZh": "认为问答材料从 Westlaw 头注提取表达并被用于训练 ROSS。", "claimIds": [], "documentIds": [] }, { "id": "matter-06:issue-copying-similarity:position-2", "issueId": "matter-06:issue-copying-similarity", "sideZh": "ROSS Intelligence", "positionZh": "未在地区法院选定作品的简易判决记录上取得有利结论，并提起中间上诉。", "claimIds": [], "documentIds": [] } ] }, { "id": "matter-06:issue-fair-use-competition", "matterId": "matter-06", "titleZh": "竞争性法律检索训练的合理使用", "questionZh": "把头注内容用于训练直接面向法律检索市场的竞争产品，是否构成合理使用？", "contextZh": "本案训练材料和产品用途具有专业法律检索背景，与通用生成式模型事实结构不完全相同。", "topic": "copyright", "status": "partly-resolved", "courtTreatmentZh": "地区法院在本案记录上否定合理使用；第三巡回法院 25-2153 上诉尚未作出终局上诉判决。", "claimIds": [ "matter-06:claim-1" ], "positions": [ { "id": "matter-06:issue-fair-use-competition:position-1", "issueId": "matter-06:issue-fair-use-competition", "sideZh": "汤森路透", "positionZh": "主张 ROSS 利用受保护编辑内容开发直接竞争的法律检索产品，市场替代性突出。", "claimIds": [], "documentIds": [] }, { "id": "matter-06:issue-fair-use-competition:position-2", "issueId": "matter-06:issue-fair-use-competition", "sideZh": "ROSS Intelligence", "positionZh": "主张训练使用应受合理使用保护，并已就不利地区法院裁判进入上诉。", "claimIds": [], "documentIds": [] } ] } ], "decisions": [ { "id": "matter-06:decision-1", "matterId": "matter-06", "proceedingId": "matter-06:proceeding-1", "eventId": "matter-06:event-1", "date": "2025-02-11", "outcomeZh": "特拉华地区法院就选定 Westlaw 头注向汤森路透作出部分简易判决，认定 ROSS 构成侵权并否定合理使用。", "holdingsZh": [ "选定 Westlaw 头注在本案范围内具有可保护的原创表达。", "ROSS 使用相关训练材料构成对选定头注的侵权。", "ROSS 在该事实记录上的合理使用抗辩不成立。" ], "reasoningZh": [ "法院逐项审查头注原创性和训练材料之间的相似性，而非保护司法意见本身。", "ROSS 系统与 Westlaw 的竞争用途及市场影响使合理使用抗辩在本案记录上不能成立。" ], "reliefZh": "就选定作品的责任问题向汤森路透作出部分简易判决；该事件未记录损害赔偿数额或独立禁令。", "scopeAndLimitsZh": "裁判限于选定头注、ROSS 的训练材料和法律检索产品事实，不能直接等同于通用生成式模型训练。", "precedentialWeight": "pending-appeal", "appealStatusZh": "第三巡回法院以 25-2153 受理中间上诉，地区法院意见目前标记为待上诉，尚无终局上诉判决。", "documentIds": [ "matter-06:document-1" ], "claimIds": [ "matter-06:claim-1" ] } ], "issueDocumentLinks": [ { "id": "issue-document-matter-06-matter-06:issue-headnote-originality-knowledge-foundation-models-fair-use", "matterId": "matter-06", "issueId": "matter-06:issue-headnote-originality", "documentId": "knowledge-foundation-models-fair-use", "relationship": "technical-context", "relevanceSummaryZh": "连接合理使用因素、输出相似度实验和技术缓解措施，适合解释训练与输出风险为何不能混为一谈。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "issue-document-matter-06-matter-06:issue-headnote-originality-knowledge-fair-learning", "matterId": "matter-06", "issueId": "matter-06:issue-headnote-originality", "documentId": "knowledge-fair-learning", "relationship": "doctrinal-context", "relevanceSummaryZh": "为训练复制属于非表达性使用的主张提供基础 doctrine，同时明确表达性模仿可能改变结论。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "supportive", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "news-context-link-matter-06-matter-06:issue-headnote-originality-news-context-ai-liability-ip-harms", "matterId": "matter-06", "issueId": "matter-06:issue-headnote-originality", "documentId": "news-context-ai-liability-ip-harms", "relationship": "news-context", "relevanceSummaryZh": "用于比较生成式 AI 供应链中训练复制、近似输出、元数据去除、商标和公开权的不同责任入口。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "issue-document-matter-06-matter-06:issue-copying-similarity-knowledge-foundation-models-fair-use", "matterId": "matter-06", "issueId": "matter-06:issue-copying-similarity", "documentId": "knowledge-foundation-models-fair-use", "relationship": "technical-context", "relevanceSummaryZh": "连接合理使用因素、输出相似度实验和技术缓解措施，适合解释训练与输出风险为何不能混为一谈。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "issue-document-matter-06-matter-06:issue-copying-similarity-knowledge-fair-learning", "matterId": "matter-06", "issueId": "matter-06:issue-copying-similarity", "documentId": "knowledge-fair-learning", "relationship": "doctrinal-context", "relevanceSummaryZh": "为训练复制属于非表达性使用的主张提供基础 doctrine，同时明确表达性模仿可能改变结论。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "supportive", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "news-context-link-matter-06-matter-06:issue-copying-similarity-news-context-ai-liability-ip-harms", "matterId": "matter-06", "issueId": "matter-06:issue-copying-similarity", "documentId": "news-context-ai-liability-ip-harms", "relationship": "news-context", "relevanceSummaryZh": "用于比较生成式 AI 供应链中训练复制、近似输出、元数据去除、商标和公开权的不同责任入口。 本次规则匹配仅使用本争点文本及其引用主张，命中 2 个确定性争点词。", "viewpoint": "mixed", "confidence": 0.9, "assignedBy": "rule" }, { "id": "issue-document-matter-06-matter-06:issue-fair-use-competition-knowledge-foundation-models-fair-use", "matterId": "matter-06", "issueId": "matter-06:issue-fair-use-competition", "documentId": "knowledge-foundation-models-fair-use", "relationship": "technical-context", "relevanceSummaryZh": "连接合理使用因素、输出相似度实验和技术缓解措施，适合解释训练与输出风险为何不能混为一谈。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "issue-document-matter-06-matter-06:issue-fair-use-competition-knowledge-fair-learning", "matterId": "matter-06", "issueId": "matter-06:issue-fair-use-competition", "documentId": "knowledge-fair-learning", "relationship": "doctrinal-context", "relevanceSummaryZh": "为训练复制属于非表达性使用的主张提供基础 doctrine，同时明确表达性模仿可能改变结论。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "supportive", "confidence": 0.8500000000000001, "assignedBy": "rule" }, { "id": "news-context-link-matter-06-matter-06:issue-fair-use-competition-news-context-ai-liability-ip-harms", "matterId": "matter-06", "issueId": "matter-06:issue-fair-use-competition", "documentId": "news-context-ai-liability-ip-harms", "relationship": "news-context", "relevanceSummaryZh": "用于比较生成式 AI 供应链中训练复制、近似输出、元数据去除、商标和公开权的不同责任入口。 本次规则匹配仅使用本争点文本及其引用主张，命中 2 个确定性争点词。", "viewpoint": "mixed", "confidence": 0.9, "assignedBy": "rule" } ] }, "issueDocuments": [ { "link": { "id": "issue-document-matter-06-matter-06:issue-headnote-originality-knowledge-foundation-models-fair-use", "matterId": "matter-06", "issueId": "matter-06:issue-headnote-originality", "documentId": "knowledge-foundation-models-fair-use", "relationship": "technical-context", "relevanceSummaryZh": "连接合理使用因素、输出相似度实验和技术缓解措施，适合解释训练与输出风险为何不能混为一谈。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "knowledge-foundation-models-fair-use", "familyId": "knowledge-family-foundation-models-fair-use", "title": "基础模型与合理使用", "originalTitle": "Foundation Models and Fair Use", "url": "https://jmlr.org/papers/v24/23-0569.html", "kind": "academic-research", "sourceId": "jmlr", "publishedAt": "2023-09", "author": "Peter Henderson; Xuechen Li; Dan Jurafsky; Tatsunori Hashimoto; Mark A. Lemley; Percy Liang", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "把合理使用 doctrine 与模型输出相似度实验放在同一框架中，认为训练和部署风险取决于输出、市场影响与缓解措施，法律和技术护栏需要共同演进。", "venue": "Journal of Machine Learning Research 24(400): 1–79", "publisher": "Journal of Machine Learning Research", "publicationState": "published", "peerReviewed": true, "paywalled": false }, "stableIdentifier": { "kind": "publisher-record", "value": "jmlr:24:400:23-0569" }, "legalTopics": [ "copyright" ], "aiRoles": [ "training-data", "model-output" ] } }, { "link": { "id": "issue-document-matter-06-matter-06:issue-headnote-originality-knowledge-fair-learning", "matterId": "matter-06", "issueId": "matter-06:issue-headnote-originality", "documentId": "knowledge-fair-learning", "relationship": "doctrinal-context", "relevanceSummaryZh": "为训练复制属于非表达性使用的主张提供基础 doctrine，同时明确表达性模仿可能改变结论。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "supportive", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "knowledge-fair-learning", "familyId": "knowledge-family-fair-learning", "title": "公平学习", "originalTitle": "Fair Learning", "url": "https://texaslawreview.org/fair-learning/", "kind": "academic-research", "sourceId": "texas-law-review", "publishedAt": "2021-02-23", "author": "Mark A. Lemley; Bryan Casey", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "主张机器为获取不受版权保护的事实、思想或功能而学习时，训练复制原则上应得到合理使用的有利评价；若训练目标是复现受保护表达，结论会更困难。", "venue": "Texas Law Review, Volume 99, Issue 4", "publisher": "Texas Law Review Association", "publicationState": "published", "peerReviewed": false, "paywalled": false }, "stableIdentifier": { "kind": "publisher-record", "value": "texas-law-review:99:4:fair-learning" }, "legalTopics": [ "copyright" ], "aiRoles": [ "training-data", "model-output" ] } }, { "link": { "id": "news-context-link-matter-06-matter-06:issue-headnote-originality-news-context-ai-liability-ip-harms", "matterId": "matter-06", "issueId": "matter-06:issue-headnote-originality", "documentId": "news-context-ai-liability-ip-harms", "relationship": "news-context", "relevanceSummaryZh": "用于比较生成式 AI 供应链中训练复制、近似输出、元数据去除、商标和公开权的不同责任入口。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "news-context-ai-liability-ip-harms", "familyId": "news-family-ai-liability-ip-harms", "title": "AI 造成知识产权损害时，责任如何分配", "originalTitle": "AI Liability for Intellectual Property Harms", "url": "https://www.lawfaremedia.org/article/ai-liability-for-intellectual-property-harms", "kind": "professional-commentary", "sourceId": "lawfare", "publishedAt": "2024-09-23", "author": "Katrina Geddes", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "文章分别分析未经许可训练、近似输出、版权管理信息、商标和形象商业利用，强调开发者、部署者与用户的责任不能用一个统一答案概括。", "venue": "Lawfare", "publisher": "Lawfare Institute", "publicationState": "published", "peerReviewed": false, "paywalled": false }, "editorialSummaryZh": "这篇评论把训练、输出和传播链条拆开，说明同一生成内容可能同时触发版权、商标与人格商业利用争议；它提供责任分配框架，不代表法院已经采纳其中结论。", "legalTopics": [ "copyright", "trademark", "personality-deepfake", "platform-liability" ], "aiRoles": [ "training-data", "model-output", "deepfake" ] } }, { "link": { "id": "issue-document-matter-06-matter-06:issue-copying-similarity-knowledge-foundation-models-fair-use", "matterId": "matter-06", "issueId": "matter-06:issue-copying-similarity", "documentId": "knowledge-foundation-models-fair-use", "relationship": "technical-context", "relevanceSummaryZh": "连接合理使用因素、输出相似度实验和技术缓解措施，适合解释训练与输出风险为何不能混为一谈。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "knowledge-foundation-models-fair-use", "familyId": "knowledge-family-foundation-models-fair-use", "title": "基础模型与合理使用", "originalTitle": "Foundation Models and Fair Use", "url": "https://jmlr.org/papers/v24/23-0569.html", "kind": "academic-research", "sourceId": "jmlr", "publishedAt": "2023-09", "author": "Peter Henderson; Xuechen Li; Dan Jurafsky; Tatsunori Hashimoto; Mark A. Lemley; Percy Liang", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "把合理使用 doctrine 与模型输出相似度实验放在同一框架中，认为训练和部署风险取决于输出、市场影响与缓解措施，法律和技术护栏需要共同演进。", "venue": "Journal of Machine Learning Research 24(400): 1–79", "publisher": "Journal of Machine Learning Research", "publicationState": "published", "peerReviewed": true, "paywalled": false }, "stableIdentifier": { "kind": "publisher-record", "value": "jmlr:24:400:23-0569" }, "legalTopics": [ "copyright" ], "aiRoles": [ "training-data", "model-output" ] } }, { "link": { "id": "issue-document-matter-06-matter-06:issue-copying-similarity-knowledge-fair-learning", "matterId": "matter-06", "issueId": "matter-06:issue-copying-similarity", "documentId": "knowledge-fair-learning", "relationship": "doctrinal-context", "relevanceSummaryZh": "为训练复制属于非表达性使用的主张提供基础 doctrine，同时明确表达性模仿可能改变结论。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "supportive", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "knowledge-fair-learning", "familyId": "knowledge-family-fair-learning", "title": "公平学习", "originalTitle": "Fair Learning", "url": "https://texaslawreview.org/fair-learning/", "kind": "academic-research", "sourceId": "texas-law-review", "publishedAt": "2021-02-23", "author": "Mark A. Lemley; Bryan Casey", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "主张机器为获取不受版权保护的事实、思想或功能而学习时，训练复制原则上应得到合理使用的有利评价；若训练目标是复现受保护表达，结论会更困难。", "venue": "Texas Law Review, Volume 99, Issue 4", "publisher": "Texas Law Review Association", "publicationState": "published", "peerReviewed": false, "paywalled": false }, "stableIdentifier": { "kind": "publisher-record", "value": "texas-law-review:99:4:fair-learning" }, "legalTopics": [ "copyright" ], "aiRoles": [ "training-data", "model-output" ] } }, { "link": { "id": "news-context-link-matter-06-matter-06:issue-copying-similarity-news-context-ai-liability-ip-harms", "matterId": "matter-06", "issueId": "matter-06:issue-copying-similarity", "documentId": "news-context-ai-liability-ip-harms", "relationship": "news-context", "relevanceSummaryZh": "用于比较生成式 AI 供应链中训练复制、近似输出、元数据去除、商标和公开权的不同责任入口。 本次规则匹配仅使用本争点文本及其引用主张，命中 2 个确定性争点词。", "viewpoint": "mixed", "confidence": 0.9, "assignedBy": "rule" }, "document": { "id": "news-context-ai-liability-ip-harms", "familyId": "news-family-ai-liability-ip-harms", "title": "AI 造成知识产权损害时，责任如何分配", "originalTitle": "AI Liability for Intellectual Property Harms", "url": "https://www.lawfaremedia.org/article/ai-liability-for-intellectual-property-harms", "kind": "professional-commentary", "sourceId": "lawfare", "publishedAt": "2024-09-23", "author": "Katrina Geddes", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "文章分别分析未经许可训练、近似输出、版权管理信息、商标和形象商业利用，强调开发者、部署者与用户的责任不能用一个统一答案概括。", "venue": "Lawfare", "publisher": "Lawfare Institute", "publicationState": "published", "peerReviewed": false, "paywalled": false }, "editorialSummaryZh": "这篇评论把训练、输出和传播链条拆开，说明同一生成内容可能同时触发版权、商标与人格商业利用争议；它提供责任分配框架，不代表法院已经采纳其中结论。", "legalTopics": [ "copyright", "trademark", "personality-deepfake", "platform-liability" ], "aiRoles": [ "training-data", "model-output", "deepfake" ] } }, { "link": { "id": "issue-document-matter-06-matter-06:issue-fair-use-competition-knowledge-foundation-models-fair-use", "matterId": "matter-06", "issueId": "matter-06:issue-fair-use-competition", "documentId": "knowledge-foundation-models-fair-use", "relationship": "technical-context", "relevanceSummaryZh": "连接合理使用因素、输出相似度实验和技术缓解措施，适合解释训练与输出风险为何不能混为一谈。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "mixed", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "knowledge-foundation-models-fair-use", "familyId": "knowledge-family-foundation-models-fair-use", "title": "基础模型与合理使用", "originalTitle": "Foundation Models and Fair Use", "url": "https://jmlr.org/papers/v24/23-0569.html", "kind": "academic-research", "sourceId": "jmlr", "publishedAt": "2023-09", "author": "Peter Henderson; Xuechen Li; Dan Jurafsky; Tatsunori Hashimoto; Mark A. Lemley; Percy Liang", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "把合理使用 doctrine 与模型输出相似度实验放在同一框架中，认为训练和部署风险取决于输出、市场影响与缓解措施，法律和技术护栏需要共同演进。", "venue": "Journal of Machine Learning Research 24(400): 1–79", "publisher": "Journal of Machine Learning Research", "publicationState": "published", "peerReviewed": true, "paywalled": false }, "stableIdentifier": { "kind": "publisher-record", "value": "jmlr:24:400:23-0569" }, "legalTopics": [ "copyright" ], "aiRoles": [ "training-data", "model-output" ] } }, { "link": { "id": "issue-document-matter-06-matter-06:issue-fair-use-competition-knowledge-fair-learning", "matterId": "matter-06", "issueId": "matter-06:issue-fair-use-competition", "documentId": "knowledge-fair-learning", "relationship": "doctrinal-context", "relevanceSummaryZh": "为训练复制属于非表达性使用的主张提供基础 doctrine，同时明确表达性模仿可能改变结论。 本次规则匹配仅使用本争点文本及其引用主张，命中 1 个确定性争点词；AI 作用分面只用于候选召回。", "viewpoint": "supportive", "confidence": 0.8500000000000001, "assignedBy": "rule" }, "document": { "id": "knowledge-fair-learning", "familyId": "knowledge-family-fair-learning", "title": "公平学习", "originalTitle": "Fair Learning", "url": "https://texaslawreview.org/fair-learning/", "kind": "academic-research", "sourceId": "texas-law-review", "publishedAt": "2021-02-23", "author": "Mark A. Lemley; Bryan Casey", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "主张机器为获取不受版权保护的事实、思想或功能而学习时，训练复制原则上应得到合理使用的有利评价；若训练目标是复现受保护表达，结论会更困难。", "venue": "Texas Law Review, Volume 99, Issue 4", "publisher": "Texas Law Review Association", "publicationState": "published", "peerReviewed": false, "paywalled": false }, "stableIdentifier": { "kind": "publisher-record", "value": "texas-law-review:99:4:fair-learning" }, "legalTopics": [ "copyright" ], "aiRoles": [ "training-data", "model-output" ] } }, { "link": { "id": "news-context-link-matter-06-matter-06:issue-fair-use-competition-news-context-ai-liability-ip-harms", "matterId": "matter-06", "issueId": "matter-06:issue-fair-use-competition", "documentId": "news-context-ai-liability-ip-harms", "relationship": "news-context", "relevanceSummaryZh": "用于比较生成式 AI 供应链中训练复制、近似输出、元数据去除、商标和公开权的不同责任入口。 本次规则匹配仅使用本争点文本及其引用主张，命中 2 个确定性争点词。", "viewpoint": "mixed", "confidence": 0.9, "assignedBy": "rule" }, "document": { "id": "news-context-ai-liability-ip-harms", "familyId": "news-family-ai-liability-ip-harms", "title": "AI 造成知识产权损害时，责任如何分配", "originalTitle": "AI Liability for Intellectual Property Harms", "url": "https://www.lawfaremedia.org/article/ai-liability-for-intellectual-property-harms", "kind": "professional-commentary", "sourceId": "lawfare", "publishedAt": "2024-09-23", "author": "Katrina Geddes", "primaryOrSecondary": "secondary", "relationship": "independent", "identifiers": {}, "language": "en", "metadata": { "abstractZh": "文章分别分析未经许可训练、近似输出、版权管理信息、商标和形象商业利用，强调开发者、部署者与用户的责任不能用一个统一答案概括。", "venue": "Lawfare", "publisher": "Lawfare Institute", "publicationState": "published", "peerReviewed": false, "paywalled": false }, "editorialSummaryZh": "这篇评论把训练、输出和传播链条拆开，说明同一生成内容可能同时触发版权、商标与人格商业利用争议；它提供责任分配框架，不代表法院已经采纳其中结论。", "legalTopics": [ "copyright", "trademark", "personality-deepfake", "platform-liability" ], "aiRoles": [ "training-data", "model-output", "deepfake" ] } } ], "sources": [ { "id": "govinfo", "slug": "govinfo", "name": "GovInfo — United States Courts Opinions", "homepage": "https://www.govinfo.gov/app/collection/USCOURTS", "type": "government", "grade": "A", "primaryOrSecondary": "primary", "score": { "traceability": 25, "corrections": 18, "ownershipTransparency": 20, "expertise": 20, "historicalAccuracy": 15, "total": 98 }, "rationale": "美国政府出版局保存法院提交的判决与命令，具备案号、法院和文书定位。", "methodologyVersion": "1.0", "reviewedAt": "2026-07-15" }, { "id": "jmlr", "slug": "jmlr", "name": "Journal of Machine Learning Research", "homepage": "https://www.jmlr.org/", "type": "academic", "primaryOrSecondary": "secondary", "rationale": "同行评审机器学习期刊，提供稳定卷期、作者、编辑和开放全文定位。", "grade": "A", "score": { "traceability": 25, "corrections": 18, "ownershipTransparency": 19, "expertise": 20, "historicalAccuracy": 14, "total": 96 }, "methodologyVersion": "1.0", "reviewedAt": "2026-07-16" }, { "id": "texas-law-review", "slug": "texas-law-review", "name": "Texas Law Review", "homepage": "https://texaslawreview.org/", "type": "academic", "primaryOrSecondary": "secondary", "rationale": "学生编辑法学评论，保留卷期、作者、引注和公开更正入口；不等同于同行评审期刊。", "grade": "A", "score": { "traceability": 24, "corrections": 16, "ownershipTransparency": 18, "expertise": 17, "historicalAccuracy": 12, "total": 87 }, "methodologyVersion": "1.0", "reviewedAt": "2026-07-16" }, { "id": "lawfare", "slug": "lawfare", "name": "Lawfare", "homepage": "https://www.lawfaremedia.org/", "type": "newsroom", "primaryOrSecondary": "secondary", "rationale": "作者和编辑准则公开，长期提供国家安全、平台治理和科技法分析；具体案件结论仍回溯官方材料。", "grade": "A", "score": { "traceability": 23, "corrections": 18, "ownershipTransparency": 19, "expertise": 19, "historicalAccuracy": 13, "total": 92 }, "methodologyVersion": "1.0", "reviewedAt": "2026-07-16", "ingestion": { "feedEnabled": false, "feedUrl": null, "verification": "not-verified", "allowedUse": "discovery-metadata-and-deep-links" } } ], "canonicalUrl": "/cases/thomson-reuters-v-ross-intelligence" }