A court in Germany says Google is responsible for wrong information from its AI.

德國一家法院表示，Google 應對其 AI 產生的錯誤資訊負責。

Two companies went to court. Google's AI said these companies did bad things. This was not true. The AI mixed information from different places and made a mistake.

有兩家公司提起訴訟。Google 的 AI 聲稱這些公司做了不法之事，但事實並非如此。AI 混淆了來自不同地方的資訊，因而產生錯誤。

Google said the AI sometimes makes mistakes. The court said this is not an excuse. The AI creates new text. Because Google owns the AI, Google is like a newspaper. A newspaper is responsible for what it prints.

Google 表示 AI 有時會出錯，但法院認為這不能作為藉口。由於 AI 會創造新文本，且 Google 擁有該 AI，因此 Google 就像報社一樣，必須對其刊登的內容負責。

Google does not agree with the court. Google wants to fight the decision. Other AI companies might have the same problem in the future. This happens because Europe and the US have different rules for technology.

Google 不認同法院的決定，並希望對此提起上訴。其他 AI 公司未來可能會面臨同樣的問題。這是因為歐洲與美國在科技法規上有所不同。

Google must take away the wrong information. Google must also pay 80% of the legal costs.

Google 必須移除錯誤資訊，並支付 80% 的法律費用。

A German court has made an initial decision that Google is responsible for false and damaging information created by its AI Overviews feature.

德國法院初步裁定，Google 需為其「AI 概覽」(AI Overviews) 功能產生的虛假且具損害性資訊負責。

The legal case began when two publishers claimed that Google's AI wrongly linked their companies to fraud and illegal business activities. The court found that the AI combined data from different sources and incorrectly attributed it to the publishers. This created serious accusations that were not present in the original texts. Consequently, this decision changes how the law views search engines, as they are no longer seen as simple messengers of other people's content.

此法律案件始於兩家出版社聲稱 Google 的 AI 錯誤地將其公司與詐欺及非法商業活動聯繫在一起。法院發現，AI 組合了來自不同來源的數據，並將其錯誤地歸因於這些出版社。這造成了原始文本中並不存在的嚴重指控。因此，這項決定改變了法律看待搜尋引擎的方式，因為它們不再被視為僅僅是他人內容的傳遞者。

The court emphasized that there is a clear difference between listing links and creating 'new and substantial statements.' Because the AI uses its own algorithms to produce original content, the court argued that Google acts as a publisher. Furthermore, the judge rejected Google's argument that user warnings about AI 'hallucinations' protect the company from liability. The court also stated that AI-generated text does not have free speech protections because it is not created by a human.

法院強調，列出連結與創造「新且實質的陳述」之間有明顯區別。由於 AI 使用其自身演算法來產生原創內容，法院認為 Google 扮演了出版社的角色。此外，法官駁回了 Google 的主張，即對用戶關於 AI 「幻覺」的警告可以免除公司的責任。法院還指出，AI 生成的文本不享有言論自由的保護，因為它並非由人類創造。

Google disagrees with the ruling and plans to appeal, claiming these were only minor errors. However, this decision could have a major impact on other AI developers, such as OpenAI and Anthropic. This case happens during a time of high tension between the European Union and US tech companies, as the EU continues to push for more control over its own digital services and imposes large fines for antitrust violations.

Google 不同意該裁定並計劃上訴，聲稱這些僅是輕微錯誤。然而，這項決定可能會對其他 AI 開發商（如 OpenAI 和 Anthropic）產生重大影響。此案發生在歐盟與美國科技公司關係緊張之際，因為歐盟持續推動對其數位服務的更多控制權，並對反壟斷違規行為處以巨額罰款。

For now, Google must delete the damaging content and pay 80% of the legal fees while it waits for the result of its appeal.

目前 Google 必須刪除具損害性的內容，並在等待上訴結果期間支付 80% 的法律費用。

A German court has issued a preliminary ruling holding Google accountable for defamatory misinformation generated by its AI Overviews feature.

德國法院已發布初步裁定，判定 Google 應為其 AI Overviews 功能產生的誹謗性錯誤資訊負責。

The legal proceedings were initiated by two publishers who alleged that Google's generative AI erroneously associated their enterprises with fraudulent activities and illicit business practices. The court's analysis determined that the AI system synthesized data from unrelated entities and attributed it to the plaintiffs, thereby creating substantive claims that were absent from the original source materials. This finding marks a departure from the traditional legal classification of search engines as mere intermediaries of third-party content.

此法律程序由兩家出版社發起，指控 Google 的生成式 AI 錯誤地將其企業與詐騙活動及非法商業行為聯繫起來。法院分析認定，AI 系統綜合了不相關實體的數據並將其歸屬於原告，從而創造了原始來源資料中不存在的實質指控。這一發現標誌著法律對搜尋引擎的分類已有所變動，不再將其僅視為第三方內容的簡單中間人。

Central to the court's reasoning is the distinction between the indexing of existing hyperlinks and the production of 'independent, new, and substantial statements.' The judiciary concluded that because the AI generates original content through proprietary algorithms, the operating entity assumes the role of a publisher. Consequently, the court dismissed Google's defense regarding user warnings about AI 'hallucinations,' asserting that such disclaimers do not absolve the provider of liability, particularly when the original sources did not make the claims in question. Furthermore, the court ruled that algorithmic outputs do not qualify for free speech protections, as they lack individual human expression.

法院論證的核心在於區分索引現有超連結與產出「獨立、全新且實質的陳述」。司法機關結論認為，由於 AI 透過專有演算法生成原創內容，營運實體因此承擔了發行者的角色。因此，法院駁回了 Google 關於 AI 「幻覺」用戶警告的辯護，主張此類免責聲明不能免除提供者的責任，尤其是在原始來源並未提出相關指控的情況下。此外，法院裁定演算法輸出不符合言論自由保護的資格，因為其缺乏人類個人的表達。

While Google has expressed disagreement with the ruling and indicated its intent to appeal—characterizing the incidents as isolated errors—the decision carries significant institutional implications. Should this precedent be upheld, it may necessitate a systemic shift in liability frameworks for other generative AI developers, such as OpenAI and Anthropic. This development occurs amidst a broader climate of regulatory tension between the European Union and United States technology firms, characterized by substantial antitrust fines and initiatives aimed at enhancing European digital sovereignty.

儘管 Google 表示不認同該裁定並指示將提起上訴——將該事件定義為孤立錯誤——但此決定具有顯著的制度影響。若此先例獲維持，可能會導致其他生成式 AI 開發商（如 OpenAI 和 Anthropic）的責任框架發生系統性轉變。這一發展正值歐盟與美國科技公司監管緊張的大環境之中，其特點在於巨額的反壟斷罰款以及旨在增強歐洲數位主權的舉措。

Google must currently remove the defamatory content and cover 80% of the legal costs pending the outcome of its appeal.

Google 目前必須刪除該誹謗內容，並在上訴結果出爐前支付 80% 的法律費用。