German Courts Treat Google's AI Answers as Its Own
A second German court has held Google responsible for false AI search output. Both rulings are first-instance, so the law is not yet settled.
Google is now answerable in German courts for what its AI search says. The Regional Court Aachen, in a 24 September 2026 injunction, applied the reasoning of the Regional Court Munich I: AI Overviews output is Google's own statement. Both rulings are first-instance, so the law is not settled.
The Weights Desk · 4 min read- The Regional Court Aachen issued a preliminary injunction on 24 September 2026 against false statements in Google's AI Mode, relying on the earlier Munich reasoning (reported by The Decoder and beck-aktuell).
- The Regional Court Munich I reportedly ruled in late May 2026 that AI Overviews are Google's own statements; Google announced an appeal but then settled, so no higher court reviewed it.
- Aachen's order is not final and was reportedly issued without hearing Google; fines of up to 250,000 euros per violation were reported.
- Binding now: for these two chambers, AI answers are the operator's content. Binding soon: appellate review. Noise: any claim that this is settled German or EU law.
- Teams shipping generative search or answer features should treat output about named individuals as publisher speech and build takedown and verification paths accordingly.
A second German court has now treated Google's AI search output as Google's own speech. The Regional Court Aachen issued a preliminary injunction on 24 September 2026, according to The Decoder and beck-aktuell, and it relied on the earlier Regional Court Munich I reasoning. The practical message for anyone shipping generative answers: the operator, not the cited third-party source, is the speaker, and ignoring a documented correction request is where exposure crystallises.
What the Aachen court ordered
The order bans further spread of three disputed text variants about a hospital chief physician, with fines of up to 250,000 euros per violation, as reported by The Decoder. The case is listed by beck-aktuell under file number 8 O 268/26 and, per the reports, turned on an AI Mode answer the court found false. This piece does not repeat the content of that answer. The order is a preliminary injunction, not a final judgment, and Google can challenge it.
The reasoning: new statements, not search results
The court said Google offers AI-generated search and is therefore responsible for the answers generated in that context, per The Decoder's account. It adopted the Munich court's description of AI output as independent, new statements assembled from third-party content rather than a neutral list of search results. If that framing holds, the familiar defence of merely relaying what a source said carries little weight, because the operator is treated as the speaker.
Why Google's appeal strategy stalled
The Regional Court Munich I ruled in late May 2026 that AI Overviews content is Google's own, in a case brought by two Munich publishers, according to The Decoder. Google announced an appeal, then settled on undisclosed terms. That means no higher court tested the reasoning, and Aachen has now reused it. The result is a consistent first-instance line without an appellate check.
What is binding now, what is soon, what is noise
Binding now: these two courts treat AI answers as the operator's content, and the Aachen injunction carries penalties unless challenged. Binding soon: a challenge to the injunction or main proceedings could produce appellate review, which would show whether the reasoning survives scrutiny. Noise: claims that this settles German or EU liability. First-instance orders do not bind other chambers, and a settlement creates no precedent at all.
Obligations for teams shipping AI answers
Prioritised: first, build a takedown path for statements about named individuals that can act within days, since The Decoder reports the injunction followed a formal demand that Google did not act on. Second, verify generated claims against the cited sources before display. Third, log generated variants so a specific text can be suppressed. Publishers whose text feeds these answers may also gain claims, an implication The Decoder raises.
Verdict: treat generated statements about real people as your own publication. The two rulings are provisional and untested on appeal, so the law may shift, but the cost of ignoring a documented correction request is now demonstrated. Details here rest on The Decoder and beck-aktuell reporting; the order text is linked in the sources, and readers should check it directly.
- Is Google liable for AI Overviews in Germany?
- Two regional courts, Munich and Aachen, have reportedly held so at first instance. Munich ended in a settlement and Aachen is a non-final preliminary injunction, so no appellate court has confirmed it.
- What did the Aachen court order?
- According to The Decoder, it banned further spread of three disputed text variants about a hospital chief physician, with fines of up to 250,000 euros per violation. Google can challenge the order.
- Why did the Aachen case arise?
- The case concerned an AI Mode answer about a physician that the court treated as false. The Decoder reports Google did not act after a formal demand. The physician is not named here, and the order itself is the place to check the details.
- Does this bind courts outside Germany?
- No. These are national first-instance decisions. They are persuasive signals about how courts may treat AI-generated answers, not precedent for other jurisdictions.
- Two German courts now hold Google liable for what its AI overviews say — The Decoder
- LG Aachen, 8 O 268/26: Haftung für falsche KI-Aussage von Google (Störerin, Chefarzt) — beck-aktuell
- Aachen preliminary-injunction order (PDF) — Institut für Informations-, Telekommunikations- und Medienrecht (ITM)