On May 28, a Munich regional court did something no court had done before. It ruled that Google is legally liable for false claims generated by its AI Overviews, the AI-written summaries that now appear above traditional search results. The court handed Google a preliminary injunction with fines of up to EUR 250,000 per violation. Google appealed on June 12. The appeal does not pause the ruling.
For brands, this changes the calculation. Your AI visibility strategy now has a legal dimension it did not have 45 days ago.
[INSIGHT] A German court just classified AI Overviews as Google's own words. That single classification dismantled two decades of search engine liability protection.
The Shield That Protected Everything
For twenty years, search engines operated behind a simple legal wall: they were intermediaries, not publishers. If a search result linked to defamatory content, the website that published it was liable. Google was not. Google was the messenger, not the message.
This doctrine, refined over decades by courts including Germany's Federal Court of Justice, rested on a practical argument. Search engines index billions of pages. They cannot proactively review every claim on every page. Requiring them to do so would make search impossible. Without it, in the BGH's own words from 2018, the internet "would be unusable for individuals due to the overwhelming flood of data."
That argument held for links. For snippets. For autocomplete suggestions. The Munich chamber looked at AI Overviews and said: this is different.
Three Arguments That Dismantled the Shield
The written reasoning, released in full on July 4, rests on three findings that together close the door on Google's intermediary defense.
The first is linguistic. Search results appear as links. AI Overviews open with a confident "Yes" and then assemble a narrative. The court noted that the overviews at issue introduced themselves with affirmations like "Yes, Verlagshaus24 is known for dubious business practices..." and then structured the answer under headings Google invented: "Characteristics of the alleged scam," "What You Can Do." None of this architecture appears in the source pages. The court held this demonstrates "an independent, content-based processing of the search results."
The second finding concerns fabrication. The overviews connected two publishing companies to third-party firms associated with scams. An affidavit from the Association of German Publishers and Booksellers confirmed no connection exists. None of the source links Google displayed alongside the overview contained that claim. The AI made it up. On that record, the court concluded the overviews contain "a separate statement generated by the AI provided by the respondent, for which the respondent must accept responsibility."
The third argument is the one that will travel furthest. The court asked what happens if AI-generated falsehoods get the same notice-and-takedown treatment as linked content. The answer: the victim has nowhere to go. The source websites did not make the false claim. Google did. If Google owes nothing beyond removing obvious violations after notification, victims of statements that require deeper verification "would then have no adequate opportunity to obtain legal protection." That gap itself, the court held, is reason enough not to extend search engine precedent to AI output.
PPC Land has the full breakdown of the three grounds, including how the court distinguished AI Overviews from two decades of BGH search engine and autocomplete precedent.
[INSIGHT] The court did not just rule against Google. It ruled that AI Overviews are a fundamentally different product from search, and the legal framework needs to reflect that.
The Front-Page Reader
During oral arguments, Google made a predictable defense. Informed users know AI-generated information should not be blindly trusted. They can check the attached links. The system is transparent about its limitations.
The court answered with an established doctrine of German press law. The front-page reader, a figure developed by the Federal Constitutional Court in 1998, takes a self-contained statement at face value. The possibility of verifying a claim through further research does not relieve the speaker of liability for making it.
Applied to AI Overviews, the court held the summary "is comprehensible in and of itself, contains a self-contained statement with independently understandable content, and makes no reference to other possible interpretations or even to any unreliability of the content." Users have no reason to doubt it. The court added a pointed observation about Google's own product logic: if users actually needed to verify every overview against every source, that "would significantly diminish the very utility of the feature cited by the respondent."
Google cannot claim the product is useful and simultaneously argue nobody should rely on it. The court caught the contradiction and used it.
What This Means for Your Brand
The ruling matters for marketing teams because it opens a door. If Google is liable for false claims its AI makes about your brand, you have legal recourse you did not have before. The calculus shifts from "we cannot control what AI Overviews say" to "Google has an obligation not to publish fabrications about us."
Three practical implications follow.
First, monitoring becomes mandatory. If an AI Overview falsely attributes a damaging claim to your brand and you do not document it, you lose the evidentiary basis for action. The publishers in the Munich case won partly because they sent attorney letters, emails, and form complaints documenting the false claims before filing. Smart brands will start capturing AI Overview outputs for their brand terms systematically.
Second, correction requests now carry legal weight. Before the ruling, asking Google to fix an AI Overview was a request with no enforcement mechanism. Now it is backed by a court precedent that gives Google a duty to verify claims against sources. The publishers notified Google through two channels. Google failed to verify. The court held that against them. Your notification creates a record.
Third, your content strategy needs a legal layer. If AI Overviews can fabricate claims about your brand by synthesizing unrelated pages, the content you publish becomes part of that synthesis pipeline. Some brands are responding by publishing more authoritative, clearly attributed content that makes misrepresentation harder. Others are publishing less directly quotable material. Neither is a complete answer. Both reflect the new reality that publishing carries AI liability exposure.
Google's AI Overviews were already draining your attribution data. Now they are adding legal risk to the balance sheet.
[INSIGHT] The liability question is not whether Google will appeal. It will. It is whether other courts and regulators will follow the Munich chamber's reasoning. The answer is almost certainly yes.
Where This Goes Next
The Munich ruling cites Article 3(3) of the EU AI Act to anchor responsibility for AI system output in national civil law. It did not apply the AI Act directly, since that regulation offers only a complaint route to market surveillance authorities. But the signal is clear: European courts are looking at the AI liability framework and building national law on top of it.
Google has appealed. The appeal does not pause the injunction. The EUR 250,000 per-violation fines stand while the appeal moves through the system. That alone changes enforcement incentives. Google can no longer afford to treat AI Overview accuracy as an aspirational metric. It is now a legal obligation.
The US path is slower but the direction is the same. State attorneys general are investigating AI companies. The FTC is expanding its deceptive practices authority to include AI-generated content. By 2027, expect US courts and regulators to cite the Munich ruling when AI systems harm consumers or misrepresent brands.
This is not a European anomaly. It is the first domino.
What to Do Right Now
If you lead marketing for a brand, here is what to do this week.
Audit your AI Overview presence. Run brand searches on Google. Capture what AI Overviews say about your company. Compare it to what you actually publish. If there are discrepancies, document them with timestamps and screenshots.
Brief your legal team. This is not a marketing problem anymore. Your general counsel needs to understand that AI Overview liability has a court precedent. The question is not whether this matters. It is whether your legal team is prepared for it.
Review your most quotable content. If you publish concise, extractable statements that could be synthesized out of context, you are feeding the same pipeline that produced the Munich falsehoods. Consider whether the visibility tradeoff is worth the exposure.
The authorization gap in AI agents and the liability gap in AI Overviews are the same problem from different angles. The technology moves faster than the legal framework. When the framework catches up, it does not just regulate. It reassigns responsibility.
The Munich court just reassigned responsibility for AI Overviews from no one to Google. Your brand is the beneficiary of that reassignment. Make sure your team knows it happened.