Google is liable for AI errors

In a final judgment dated May 28, 2026 (case no. 26 O 869/26, n.rk), the Munich I Regional Court, in preliminary injunction proceedings, prohibited Google from using its “AI Overview” feature to disseminate several specific, defamatory, and false factual claims about two Munich-based publishers. In the overview displayed, the AI had attributed fraudulent activities (including “scams,” “subscription traps,” and dubious business models) to these publishers, even though these allegations were in fact directed at other companies.

AI responses are responses from Google

In this case, the court had to address the fundamental question of whether the liability exemptions established for traditional search results also apply to AI-generated overviews. Under these liability exemptions—which provide that direct liability arises only in exceptional cases involving obvious violations and, otherwise, generally only after specific notice, since only third-party content is displayed—the court classified the AI summary as Google’s own content and not merely as a display of third-party search results. Unlike conventional search result lists with snippets, the AI aggregates multiple sources into a standalone, self-contained continuous text. This result interacts with the user’s query and is perceived by the user as an answer from Google. Against this backdrop, the court held that the liability exemptions developed for traditional search results are not applicable here. Rather, the standard legal standards governing liability for false factual claims apply directly. Finally, even a purported disclaimer such as “generated by AI” is irrelevant and does not alter the attribution to Google.

General right of personality prevails over special provisions

The legal basis for the court’s decision was the general provisions protecting the general right of personality—namely, Sections 1004 and 823(1) of the German Civil Code (BGB) in conjunction with Article 1(1) and Article 2(1) of the German Constitutional Law (GG)—which, pursuant to Article 14(1) and Article 19(3) of the GG, also apply to legal entities as so-called corporate personality rights, e.g., for companies or other institutions.

If the statement is a false assertion of fact, unlawful infringements give rise to claims for injunctive relief. In the court’s view, neither the GDPR—which in any case does not apply to purely legal entities—nor the Digital Services Act (DSA) supersede national civil law claims for injunctive relief. With regard to the latter, Article 6(4) of the DSA expressly leaves court orders to cease and remedy infringements unaffected.

Practical consequences

As a result of the ruling, Google must not only remove the specific content that was the subject of the complaint but also implement organizational measures to ensure that the AI feature no longer generates comparable false claims about the plaintiffs in the future. Beyond the specific facts of this case, the decision is likely to lower the threshold for both companies and individuals to take action—including through preliminary injunctions—against AI-generated summaries that damage their reputation.

Outlook

The decision makes it clear that the use of AI is not exempt from liability and thus sends an important signal that provides clarity while also building trust. After all, the skepticism that the use of AI faces in many quarters can only be countered by clear and transparent liability rules. This is also in the best interests of all parties; those who expand their scope of action through AI are also liable for the associated risks. In the long term, Regulation (EU) 2024/1689 establishing harmonized rules for artificial intelligence (AI Regulation) will not change this, as it will only provide for the possibility of filing a complaint with the competent market surveillance authority, but will apply in addition to and “without prejudice” to the legal remedies already provided for in Union law and in the national laws of the Member States (Recital 170 of the AI Regulation).