Two German Courts Treat Google's AI Answers as Its Own Statements

The answer is the statement

The distinction both courts draw is between pointing and speaking. A search result points to someone else's page. An AI Overview, or an AI Mode answer, combines material from several pages and writes a new text. Munich treated that new text as a statement attributable to Google, and Aachen, citing Munich directly, treated AI Mode the same way: "The respondent provides AI-generated search and is therefore also responsible for the answers generated in this context."

That classification changes the starting point of the legal analysis. Rather than treating an AI-generated answer as a neutral reference to third-party material, the courts considered Google responsible for statements its systems generated. Whether a particular answer is unlawful still depends on its content and the applicable legal standards. The reasoning makes it harder for Google to rely on the distinction between hosting or linking to third-party information and generating its own factual assertions. It does not mean every legal rule governing traditional publishers automatically applies to AI answers. German commentary frames the responsibility as Störerhaftung — disturber liability, supporting orders against the specific false claims — which is narrower than a general publisher regime.

Aachen is the harder case to dismiss

Munich involved publishers falsely linked to fraud schemes. Aachen involved a named professional and a fabricated sexual-assault narrative. The AI took a real case, one brought against a different former professor, and attached it to the doctor, then filled in a dismissal and a legal aftermath that never happened to him. The doctor himself was accused of nothing by any court.

The complaint record is part of what makes the order difficult to wave away as a court misunderstanding the technology. The doctor explained the mix-up in detail in late August. Google replied in mid-September that it would take no action. A formal legal demand followed the same day. According to the court, the texts were still live. The judges also made a practical point about capability: Google could compare its AI output against the source websites it drew on, and had it put the doctor's objections to its own system, the system would likely have flagged the errors itself.

The order itself is narrow. It bars three specific disputed text variants, it was issued on an emergency basis without hearing Google first, and Google can challenge it. The €250,000 figure is the maximum fine for breaching the order, not a penalty already imposed.

Two Regional Courts Agree. No Higher Court Has Ruled.

The Munich reasoning has now been applied twice, and tested at a higher level zero times. Google announced an appeal in the Munich case, but the dispute subsequently settled on undisclosed terms. As a result, no higher court reviewed the reasoning — and, according to German legal commentary, the settlement left the Munich judgment itself without operative effect. The settlement's motivations are not established, and Aachen's preliminary injunction remains open to challenge.

That gap cuts in both directions, and readers should hold both halves. Two regional courts agreeing is a real signal about how German judges currently see AI answers. It is not a settled rule of German law, and it says nothing directly about other jurisdictions.

Nor are Munich and Aachen the only German lower-court proceedings on the question. Frankfurt Regional Court considered liability for AI Overviews in September 2025: it accepted that liability was possible in principle, but rejected the claim before it because it did not consider the disputed statement false in context. That reinforces the distinction at the centre of all three cases — attributing an AI statement to Google is one step; establishing that a particular statement is unlawful is another.

The wider consequence is the one neither case decided. The Decoder suggests the reasoning could eventually influence disputes over the use of publishers' content in AI-generated answers. Neither the Munich nor the Aachen proceeding decided whether source publishers are owed compensation, and the legal tests for such claims would be distinct from those governing false statements. It is still the reason the liability question is worth watching beyond the individual false claims: a final ruling on the "own statements" point would reach the economics of AI search, not just its accuracy.

For now, the practical position in Germany is concrete enough. A person harmed by a false AI answer has two court decisions to cite, a complaint route that in Aachen only worked once lawyers and a court were involved, and a line of reasoning no higher court has yet endorsed, narrowed or rejected.

Sources

[1] The Decoder, "Two German courts now hold Google liable for what its AI overviews say", 18 October 2026 — primary reporting, read in full: Munich late-May ruling and its own-statements classification; announced appeal ending in settlement, terms unknown, no comment from either side; Aachen decision dated 24 September 2026 expressly relying on Munich; the doctor's role and tenure, the 28 August alert, the misattributed case and invented legal aftermath; the late-August email complaint, mid-September refusal and unanswered formal demand; the bar on three disputed text variants, emergency basis without hearing Google, fines up to €250,000 per violation, Google's right to challenge; the quoted court reasoning on independent new statements and on checking output against sources; the publisher-claims consequence framed as a possible next stepRead source
[2] LHR Rechtsanwälte Köln, legal commentary on the Aachen order, October 2026 — confirms the Frankfurt judgment of 10 September 2025 (liability prerequisites affirmed in principle, claim denied on the facts), the Munich judgment of 28 May 2026 (case 26 O 869/26) and that the parties ended the Munich dispute by settlement in the appeal proceedings, leaving the judgment without operative effect, and the Aachen order of 24 September 2026 (case 8 O 268/26), issued without an oral hearing and open to objectionRead source
[3] anwalt.de, practitioner FAQ on the Aachen order, October 2026 — corroborates both case numbers and dates, the Störerhaftung framing (liability as disturber for injunctive relief), and that the question has not been decided by a highest courtRead source

THE ALFW VERDICT

Two German regional courts have treated AI-generated search answers as statements attributable to Google rather than neutral links to third-party content. In the cases before them, that reasoning supported action over false factual claims — including a preliminary injunction against three AI Mode texts that wrongly associated a doctor with another person's sexual-assault case. The position remains unsettled: the Munich dispute ended in settlement without appellate review, and the Aachen order can still be challenged. What the decisions show is an emerging approach to responsibility for AI-generated answers, not a general rule that every AI response creates liability, and not a ruling that source publishers are entitled to compensation. The next development that matters is whether a higher court endorses, narrows or rejects the reasoning.