Shanghai Court Rules User Liable for AI-Generated Copyright Infringement in 'Battle Through the Heavens' Case

Thumbnail Image

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

A Shanghai court ruled that a user infringed copyright by using an AI platform's LoRA training feature to generate and share images of the 'Battle Through the Heavens' character Medusa, ordering the user to pay 50,000 RMB in damages. The AI platform was not held liable, having complied with takedown obligations.[AI generated]

Why's our monitor labelling this an incident or hazard?

The event explicitly involves AI systems (large AI models and LoRA models) used to generate images that infringe on copyrighted character images. The infringement of copyright constitutes a violation of intellectual property rights, which is a form of harm under the AI Incident category. The court ruling confirms that harm has occurred due to the AI system's use. Therefore, this qualifies as an AI Incident because the AI system's use directly led to a breach of intellectual property rights.[AI generated]
Industries
Media, social platforms, and marketing

Affected stakeholders
Business

Harm types
Economic/Property

Severity
AI incident

AI system task:
Content generation


Articles about this incident or hazard

Thumbnail Image

【AI】上海首例人工智能大模型著作權侵權案宣判,用戶被判賠5萬元

2025-11-04
ET Net
Why's our monitor labelling this an incident or hazard?
The event explicitly involves AI systems (large AI models and LoRA models) used to generate images that infringe on copyrighted character images. The infringement of copyright constitutes a violation of intellectual property rights, which is a form of harm under the AI Incident category. The court ruling confirms that harm has occurred due to the AI system's use. Therefore, this qualifies as an AI Incident because the AI system's use directly led to a breach of intellectual property rights.
Thumbnail Image

上海首例AI大模型侵权案,《斗破苍穹》美杜莎被"喂"给AI大模型 法院判侵权

2025-11-04
扬子网(扬子晚报)
Why's our monitor labelling this an incident or hazard?
The event explicitly involves an AI system (an AI image generation platform using large models and LoRA training) whose use directly led to copyright infringement harm. The unauthorized use of copyrighted images to train AI models and generate derivative content violates the copyright holder's rights, which is a breach of intellectual property rights under the framework. The court ruling confirms the harm has materialized and the AI system's role is pivotal in causing this harm. Hence, this is an AI Incident rather than a hazard or complementary information.
Thumbnail Image

《斗破苍穹》被AI"抄袭":用户判赔5万,大模型公司免责-证券之星

2025-11-05
wap.stockstar.com
Why's our monitor labelling this an incident or hazard?
The event explicitly involves an AI system (a large generative model with fine-tuning capabilities) used by a user to reproduce copyrighted anime character images without authorization, constituting copyright infringement. The harm is realized as the court ordered the user to pay damages for violating copyright law, which is a breach of intellectual property rights. The AI platform's role was indirect and mitigated by its compliance with takedown and filtering obligations, so the primary harm stems from the user's AI misuse. Therefore, this qualifies as an AI Incident due to direct involvement of AI in causing intellectual property rights violations.
Thumbnail Image

21世纪经济报道记者肖潇 报道 AI版权之争尚未停歇,但边界正一点点变得清晰。11月3日,上海金山区人民法院宣判了上海首例AI著作权侵权。该案的涉事方分别为《斗破苍穹》系列动漫的版权方、国内一家头部AI平台,以及该平台用户。 据上海高院介绍,被告用户擅自截取《斗破苍穹》中角色"美杜莎"的动漫形象,用于微调大模型,并将模型发布在平台社区供他人使用。上海法院认定,这一行为构成侵犯著作权,用户需赔偿原告5万元。 而AI公司虽然提供了大模型技术并设有"动漫专区",但在收到起诉状后,及时下架美杜莎模型,更新关键词过滤。法院因此认为,公司已履行"通知 -- 删除"义务,无需承担连带责任。 这起案件,为国内AI版权问题提供了新的判例参考。当越来越多AI产品拥有UGC社区和微调功能,能否及时响应侵权投诉、有没有在产品中提示侵权风险,逐渐成为关键合规指标。 2009年,网文作家天蚕土豆开始在起点中文网连载《斗破苍穹》。这部长达532万字的玄幻小说很快席卷读者圈,成为起点首部点击量破亿的作品。2017年,腾讯视频与阅文集团联合出品《斗破苍穹》动画,此后小说IP一路延展出手游、真人电视剧等完整的"斗破宇宙"。 公开资料显示,《斗破苍穹》小说的复制权、信息网络传播权等著作财产权属于上海玄霆娱乐信息科技有限公司,该公司属于腾讯旗下的阅文集团。阅文集团在2025年中期业绩报告中特别提到,《斗破苍穹》动画上半年登顶腾讯视频付费榜,商业价值不言而喻。 在本次案件中,被《斗破苍穹》版权方诉至法庭的有两方:提供文生图大模型的国内AI公司,以及平台用户李某。 据上海法院披露,用户李某截取了《斗破苍穹》中女主角"美杜莎"的二十余张动漫形象,用AI平台的"训练LoRA"功能微调出美杜莎大模型,并通过该平台机审后公开发布。此后,其他用户只需输入提示词,就能生成与美杜莎形象相似的图片。 上海法院最终判定,李某的行为确实构成侵权,判赔经济损失3万元和维权开支2万元。 据法院解释,李某以商业使用为目的,在训练、发布、使用过程中再现了《斗破苍穹》美杜莎的原创性表达,并且让公众能够通过AI生成与原角色相似的图像。这一行为满足著作权侵权中"接触"和"实质性相似"两项标准,侵犯了原告的复制权和信息网络传播权。 不过,上海法院也指出,由生成式人工智能直接生成的图片,并非著作权法意义上的美术作品。也就是说美杜莎大模型和AIGC图片不构成"作品",所以李某的行为未侵犯改编权。 值得一提的是,《斗破苍穹》的"AI再创作"早已在网上泛滥。今年以来,抖音上流行用AI重制《斗破苍穹年番》中缺失的剧情,不少视频长达3分钟,点赞过万。21记者于11月4日搜索,仍看到不少AIGC自媒体团队发布《斗破苍穹》AI视频,评论区不乏粉丝留言:"终于找到了正版""我宣布你才是官方"。 除了用户,AI平台要不要为侵权行为担责,一直也是业界争论的焦点。在这起案件中,上海法院认定AI平台不构成侵权,理由主要有两点: 从技术与服务特征看,AI平台并未直接参与"美杜莎"模型的素材截取、训练、发布和使用。换句话说,《斗破苍穹》被当作训练素材是独立的用户行为。 另一方面,法院认为AI平台已尽到合理注意义务。按照我国民法典的具体规定,网络平台要对明显的侵权行为采取必要措施,否则要承担连带责,这也是行业内常说的"红旗原则"。 根据上海高院的披露,本次案件的AI公司设置了投诉举报机制和发布审核机制,并在接到起诉状后,第一时间下架所有美杜莎模型,更新平台审核的关键词,并将情况同步给海外平台。法院因此认定,公司主观上无明显过错,客观上也履行了必要的制止与转通知义务。 这一判决,与2024年杭州互联网法院审理的"奥特曼AI侵权案"有相似之处。当时法院同样认为:用户自行上传训练素材并独立微调模型,且无证据显示平台与用户共同实施侵权行为,因此AI公司不构成直接侵权。 长期关注AIGC的浙江垦丁(北京)律师事务所律师程念向21记者表示,互联网平台责任认定过去已有成熟体系,但AI平台的出现引入了UGC社区、LoRA微调等新模式,因此需要重新思考责任问题。如今一个核心分歧在于:AI平台究竟是直接侵权的"内容提供者",还是单纯的"技术提供者"? "要回答这个问题,得看平台的具体行为。"程念解释说,比如平台若主动开设"角色模型"或"某某角色专区",甚至在设计上对用户生成特定形象起到明显导向作用,那么平台的角色就可能从技术服务者,转变为内容生产者。 在杭州"奥特曼AI侵权案"中,法官沙丽也有类似观点。她认为,由于AI生成和LoRA微调的便捷性,用户创作的内容往往可被他人反复使用,侵权扩散的后果"已相当明显",AI平台应当"预见到侵权行为发生的可能性",承担更高注意义务。 如今,这种"注意义务"已成为AIGC司法领域反复强调的关键词。但AI平台具体该如何落实,仍需通过一个个具体案件积累。 程念建议,目前AI公司至少应从三方面完善:第一,保持侵权投诉渠道畅通,确保响应高效,这在本案的免责证据中至关重要;第二,主动升级审核与风控体系,特别是对模型名称、标签、示例图等进行关键词和图像识别筛查;第三,完善用户协议与指引。尤其在用户使用训练功能时,应以显著方式提醒其需取得素材授权,并明确告知潜在的侵权风险。

2025-11-05
证券之星
Why's our monitor labelling this an incident or hazard?
The event involves an AI system (a large image generation model with fine-tuning capabilities) whose use by a user directly led to copyright infringement harm, specifically violation of intellectual property rights. The court ruling confirms the infringement and damages, constituting realized harm caused by the AI system's use. The AI platform's role and response are also detailed, but it was not held liable. This fits the definition of an AI Incident because the AI system's use directly led to a violation of intellectual property rights, a breach of applicable law protecting such rights. The article is not merely about AI policy or general AI developments, but about a concrete legal incident involving AI use and harm.
Thumbnail Image

合规科技 《斗破苍穹》被AI"抄袭":用户判赔5万,大模型公司免责

2025-11-05
21jingji.com
Why's our monitor labelling this an incident or hazard?
The event explicitly involves an AI system (a large generative AI model with fine-tuning capabilities) used by a user to generate images infringing on copyrighted anime characters. The court ruling confirms that the user's use of the AI system directly caused copyright infringement harm, satisfying the criteria for an AI Incident under violations of intellectual property rights. The AI platform's role was limited and it complied with legal obligations, so it is not liable, but the incident centers on the AI system's use causing harm. Hence, this is classified as an AI Incident.
Thumbnail Image

上海首例人工智能大模型著作权侵权案宣判,用户被判赔偿5万元

2025-11-03
k.sina.com.cn
Why's our monitor labelling this an incident or hazard?
The event involves the use of an AI system (large AI model and LoRA models) to generate images that infringe on copyrighted character images, leading to a legal ruling and compensation for copyright infringement. The AI system's use directly caused harm by violating intellectual property rights, fulfilling the criteria for an AI Incident. The platform's role as a service provider and its compliance with takedown and notification obligations do not negate the fact that the AI-generated content caused harm. Therefore, this is classified as an AI Incident due to realized harm involving copyright infringement through AI-generated content.