# 德国两家法院先后裁定 Google 须为其 AI Overview 内容负责
> 原标题:Two German courts now hold Google liable for what its AI overviews say
- 来源:The Decoder:AI News
- 发布时间:2026-10-18T10:52:39.000Z
- AIHOT:https://aihot.news/items/bbr25r9mbi8zbbtmu2uiati7f
- 原文:https://the-decoder.com/two-german-courts-now-hold-google-liable-for-what-its-ai-overviews-say/
## 摘要
慕尼黑地区法院 I 五月底裁定 Google 的 AI 搜索摘要属于公司自己的陈述而非中立转述,Google 须为错误内容担责;Google 公开宣布上诉后实际以和解结案,未获上级法院裁决。
## 推荐理由
两起德国法院判例指向同一法律逻辑,文章还分析了 Google 以和解回避判例对 AI 搜索商业模式的潜在影响。
## 正文 · 中文译文
**Google曾试图通过上诉推翻慕尼黑一项关于AI搜索摘要责任的里程碑式裁决,随后悄然达成和解。如今,第二家法院采用了同样的推理。Google的AI回答被视为该公司自身的内容,Google需对其负责。**
5月下旬,慕尼黑地区法院第一庭在一桩[备受关注的案件](https://the-decoder.com/landmark-german-ruling-declares-googles-ai-overviews-are-googles-own-words-and-makes-it-liable-for-false-answers/)中裁定,Google对其AI生成的搜索摘要内容负有直接责任。Google的AI曾错误地将两家慕尼黑出版商与欺诈计划关联。法官将AI概览归类为Google自身的陈述,而非单纯的搜索结果,这意味着该公司本身必须为其负责。
Google无法再躲在传递第三方信息的中立中间人角色背后,而必须为其AI生成的不法陈述承担责任。对于一个不断生成新答案的搜索功能而言,这是一种难以量化的法律风险。如果这一观点成立,可能会动摇德国AI搜索的整个商业模式,并影响其他国家的相关讨论。
搜索引用与原创内容之间的区分还可能产生超出虚假事实陈述之外的后果。如果AI回答被视为Google自身的内容,那么Google用来生成这些回答的文本所属的出版商,最终可能会主张自己的权利。这并非慕尼黑案件的一部分,但却是合乎逻辑的下一步。
据THE DECODER了解,[Google公开宣布的上诉](https://the-decoder.com/google-appeals-ruling-that-made-it-directly-liable-for-ai-generated-search-overview-content/)从未走到裁决阶段。案件最终以和解告终,这意味着没有任何上级法院对慕尼黑的推理作出评判。为何达成和解以及是否涉及金钱往来尚不清楚。Google和受影响的出版商均拒绝置评。法院也拒绝就其裁决在具体案件之外的更广泛意义发表意见。
## 和解无法抹去慕尼黑法院的推理
不过,慕尼黑法官的法律推理并未消亡。在[2026年9月24日的一项裁决中,亚琛地区法院](https://www.itm.nrw/wp-content/uploads/2026/10/KM_C450i26100808190.pdf)禁止Google传播某些虚假陈述,并明确援引了慕尼黑裁决的推理。Google可以对该禁令提出异议。该裁决尚未最终生效。
在这起新案件中,亚琛的一位医生对Google“AI模式”搜索功能中的严重虚假指控进行了反击。这位医生自2010年起至2026年6月下旬担任神经放射学首席医师和教授。8月28日,一位同事提醒他,Google的AI在解释其离职时,援引了一起据称针对其秘书的性侵事件。当用户使用“停职”“解雇”和“法庭程序”等词进行进一步搜索时,AI模式返回了详细的多段捏造内容。它甚至编造了完整的法律后续。
这些指控和程序均与该医生毫无关系。AI将涉及另一位前大学教授的事件安在了他身上,亚琛法院可以从自己的记录中确认这一点,因为刑事案件实际上是对另一个人提起的。[此类AI错误归因的案例](https://the-decoder.com/microsofts-copilot-falsely-accuses-court-reporter-of-crimes-he-covered/)此前已有[记录](https://the-decoder.com/chatgpts-bizarre-child-murder-claims-about-arve-hjalmar-holmen-leave-some-questions-unresolved/)。
## Google拒绝删除内容,面临高额罚款
尽管医生在8月底详细描述了这一混淆情况,并通过电子邮件要求谷歌删除相关内容,但该公司在9月中旬表示不会采取任何行动。同日发出的正式法律要求同样没有下文。谷歌确认已收到该要求,但据法院称,这些由AI生成的虚假文本当时仍在线。
随后,亚琛地区法院禁止进一步传播三个有争议的文本版本。由于这些虚假陈述可能严重损害医生的职业声誉,法院在未听取谷歌意见的情况下采取了紧急行动。如果谷歌违反该命令,每次违规将面临最高25万欧元的罚款。
亚琛裁决最重要的部分在于它与慕尼黑推理的高度一致,这表明德国法院正在形成一致的法律观点。谷歌利用其AI生成自己的陈述,而非仅仅让第三方信息可被找到。法院写道:“被申请人提供AI生成的搜索,因此也应对在此背景下生成的答案负责。”
亚琛甚至直接引用了慕尼黑的裁决,指出谷歌的AI基于对多个第三方网站内容的评估和组合,生成“独立的、新的、实质性的陈述”。
法院表示,谷歌可以核查此类陈述,例如将其AI输出与其使用的来源网站进行比对。亚琛的法官还补充说,如果谷歌让其自家AI面对医生的异议,系统本会自行发现这些错误。
## 影响远超个别虚假陈述
两起案件都引发了人们对谷歌针对其AI搜索功能所采取长期法律策略的疑问。如果德国法院继续将该公司归类为虚假陈述的发布者,而谷歌继续通过很可能是财务和解的方式规避判例,就会形成一种反常的激励。
对于专业律师事务所和受影响方而言,这可能变成一种有利可图的商业模式。他们可以用有针对性的查询探测AI搜索,直到语言模型拼凑出虚假、可能具有诽谤性的陈述,然后申请禁令并协商和解。这些系统很容易被操纵。
当然,谷歌有能力以这种方式让个别案件消失多年,因为只有至少一方将法律程序推进到底,才会形成具有约束力的判例。但谷歌的动机很可能远不止对个别虚假陈述的责任。如果最高法院裁定AI答案算作原创内容,那么为这些答案提供文本的发布者也可以主张自己的权利。在最极端的情况下,谷歌将不得不就其来源背后的创作者为每个AI生成的答案进行补偿。这将彻底颠覆AI搜索的经济模式。
### 没有炒作的AI新闻——由人类精选
订阅 THE DECODER,享受无广告阅读、每周AI新闻通讯、每年六期独家“AI Radar”前沿报告、完整档案访问权限以及评论区访问权限。
## 正文 · 原文
**Google tried to overturn a landmark Munich ruling on liability for AI search summaries by appealing, then quietly settled. Now a second court has applied the same reasoning. Google's AI answers count as the company's own content, and Google is responsible for them.**
In late May, the Munich Regional Court I ruled in a [widely watched case](https://the-decoder.com/landmark-german-ruling-declares-googles-ai-overviews-are-googles-own-words-and-makes-it-liable-for-false-answers/) that Google is directly liable for the content of its AI-generated search summaries. Google's AI had falsely linked two Munich-based publishers to fraud schemes. The judges classified AI Overviews not as mere search results but as Google's own statements, meaning the company itself has to answer for them.
Instead of hiding behind the role of a neutral middleman passing along third-party information, Google would have to own the unlawful statements its AI produces. For a search feature that constantly generates new answers, that's a legal risk that's hard to quantify. If this view holds up, it could call into question the entire business model of AI search in Germany and shape the debate in other countries too.
The distinction between a search reference and original content could also have consequences beyond false factual claims. If AI answers count as Google's own content, the publishers whose texts Google uses to generate those answers might eventually assert their own claims. That wasn't part of the Munich case, but it's a logical next step.
As THE DECODER has learned, the [appeal Google publicly announced](https://the-decoder.com/google-appeals-ruling-that-made-it-directly-liable-for-ai-generated-search-overview-content/) was never carried through to a decision. The case ended in a settlement instead, meaning no higher court ever weighed in on the Munich reasoning. Why the settlement was reached and whether money changed hands is unknown. Neither Google nor the affected publisher would comment. The court also declined to weigh in on the broader significance of its own ruling beyond this specific case.
## A settlement can't erase the Munich court's reasoning
The Munich judges' legal reasoning isn't dead, though. In a [decision dated September 24, 2026, the Aachen Regional Court](https://www.itm.nrw/wp-content/uploads/2026/10/KM_C450i26100808190.pdf) barred Google from spreading certain false statements and explicitly relied on the Munich ruling's reasoning. Google can challenge the injunction. The decision isn't final yet.
In this new case, a physician in Aachen fought back against serious false claims in Google's "AI Mode" search feature. The doctor had worked as chief physician and professor of neuroradiology from 2010 until late June 2026. On August 28, a colleague alerted him that Google's AI was explaining his departure by citing an alleged sexual assault against his secretary. When users ran additional searches using terms like "suspension," "termination," and "court proceedings," AI Mode returned detailed, multi-paragraph fabrications. It even invented a complete legal aftermath.
None of the allegations or proceedings had anything to do with the doctor. The AI had taken events involving a different former university professor and pinned them on him, something the Aachen court could confirm from its own records since the criminal case had actually been brought against someone else. [Cases of AI misattribution like this](https://the-decoder.com/microsofts-copilot-falsely-accuses-court-reporter-of-crimes-he-covered/) have been [documented before](https://the-decoder.com/chatgpts-bizarre-child-murder-claims-about-arve-hjalmar-holmen-leave-some-questions-unresolved/).
## Google refuses to delete the content and risks steep fines
Even though the doctor described the mix-up in detail in late August and asked Google by email to remove the content, the company said in mid-September that it would take no action. A formal legal demand sent the same day also went nowhere. Google confirmed receiving it, but according to the court, the false AI-generated texts were still live.
The Aachen Regional Court then banned further spread of the three disputed text variants. Because the false claims could seriously damage the doctor's professional reputation, the court acted on an emergency basis without hearing Google first. If Google violates the order, it faces fines of up to 250,000 euros per violation.
The most significant part of the Aachen decision is how closely it tracks the Munich reasoning, suggesting a consistent legal view is forming across German courts. Google uses its AI to produce its own statements rather than simply making third-party information findable. "The respondent provides AI-generated search and is therefore also responsible for the answers generated in this context," the court wrote.
Aachen even cites the Munich ruling directly, noting that Google's AI produces "independent, new, and substantive statements based on an evaluation and combination of content from various third-party websites."
Google could check such statements, the court said, for example, by comparing AI output against the source websites it used. The Aachen judges added that had Google confronted its own AI with the doctor's objections, the system would have caught the errors itself.
## The stakes go far beyond individual false claims
Both cases raise the question of what long-term legal strategy Google is pursuing with its AI search features. If German courts keep classifying the company as a publisher of false statements and Google keeps dodging precedent through what are likely financial settlements, that creates a perverse incentive.
For specialized law firms and affected parties, it could turn into a lucrative business model. They could probe AI search with targeted queries until the language model stitches together false, potentially defamatory statements, then file for an injunction and negotiate a settlement. These systems are easy to manipulate.
Of course, Google can afford to make individual cases disappear this way for years, since binding precedent only happens when at least one side pushes the legal process all the way to the end. But Google's motivation likely goes well beyond liability for individual false claims. If a top court were to rule that AI answers count as original content, the publishers whose texts feed those answers could assert their own claims too. In the most extreme scenario, Google would have to compensate the creators behind its sources for every AI-generated answer. That would upend the economics of AI search entirely.
### AI News Without the Hype – Curated by Humans
Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.