2026年5月28日，德国慕尼黑地方法院就一起具有里程碑意义的案件作出裁定：谷歌必须为其AI搜索摘要（AI Overviews）功能生成的虚假诽谤性内容直接承担法律责任。
一、AI生成内容属于”自身陈述”
法院明确认定，谷歌AI概览并非简单展示第三方网页内容，而是用自身措辞重新生成独立陈述，这与传统搜索引擎存在本质区别。法院指出，传统搜索引擎仅指向外部网站，属于”间接侵权者”；而AI概览”评估、整合、重新编写并结构化信息，生成独立、全新且实质性的陈述”，属于”谷歌自身的内容”。
二、打破搜索引擎免责惯例
德国联邦最高法院此前判例认为，搜索引擎运营商仅作为间接侵权者承担责任，因其仅让第三方内容变得可被查找。但慕尼黑法院认为，这一逻辑不适用于AI概览功能。法院强调，谷歌开发了该人工智能，并将其提供给用户，因此对AI生成内容负有完全主导权，理应作为直接侵权主体承担法律责任。
三、驳回谷歌抗辩理由
谷歌辩称用户可以自行点开链接核实真伪，法院断然驳回：AI摘要是完整定论式内容，普通用户没有义务逐一核验溯源，就像新闻标题误导读者，媒体必须担责，无需用户读完正文追责。谷歌还辩称AI本身存在误差、内容不可信，法院同样驳回，指出谷歌一边宣传AI摘要高效可靠，一边用AI缺陷推卸责任，逻辑自相矛盾。
判决结果与后续
法院发布临时禁令，禁止谷歌通过AI概览继续传播针对两家出版商的虚假言论。违规处罚依据德国民事诉讼法第890条：单次最高罚25万欧元；情节严重时，企业负责人最高可被拘留6个月。目前该判决仅为临时禁令、非终审判决，谷歌已表示正在评估上诉方案。
对比中国判决
2026年1月，杭州互联网法院审理中国首例AI幻觉侵权案时，驳回原告诉讼请求。法院认为，AI不具有民事主体资格，不能作出意思表示；AI自行生成的”赔偿承诺”也不能视为服务提供者的意思表示；案涉生成式人工智能属于服务而非产品，不应适用无过错责任原则。
1. AI-Generated Content Constitutes “Google’s Own Statements”
The court clearly determined that Google’s AI Overview does not simply display third-party web content but generates independent statements using its own wording—this represents a fundamental distinction from traditional search engines. The court noted that traditional search engines merely point to external websites, classifying their operators as “indirect infringers.” However, AI Overview “evaluates, combines, rewrites, and structures information into new statements,” constituting “Google’s own content.”
2. Breaking Search Engine Immunity Precedent
Previous rulings by the German Federal Court of Justice held that search engine operators bear only indirect infringement liability since they merely make third-party content discoverable. However, the Munich court determined this logic does not apply to AI Overview functionality. The court emphasized that Google developed and operates this AI system, maintaining complete control over its algorithms and content output, thus bearing direct responsibility as the primary tortfeasor.
3. Rejection of Google’s Defense Arguments
Google argued that users could verify information by clicking on links. The court firmly rejected this: AI summaries present complete, definitive conclusions, and ordinary users have no obligation to verify each detail—much like how media outlets must bear responsibility for misleading headlines without requiring readers to finish the full article. Google also claimed AI inherently contains errors and its content is unreliable. The court dismissed this argument as logically inconsistent, noting Google promotes AI Overview as efficient and reliable while simultaneously using AI defects to evade responsibility.