Can Google’s AI Overview get your business wrong? Yes — and a court just ruled Google is liable for it.
In May 2026, a German court did something no court had clearly done before: it held Google directly responsiblefor false statements its AI Overviews generated about real businesses. Not as a middleman pointing at someone else’s content — as the author of the claim.
If an AI engine can confidently describe your business as a “scam” with no basis in its own sources, the question stops being academic. Do you even know what AI engines are saying about you right now?
What the court actually ruled
The Regional Court of Munich (LG Munich I, case 26 O 869/26) issued an injunction on 28 May 2026stopping Google from repeating false claims about two publishers. Its AI Overviews had tied their names to scams, subscription traps and “dubious business practices” — claims that appeared in none of the sources the AI cited.
The court’s reasoning matters more than the result:
AI summaries are the operator’s own words
Unlike a list of search links, an AI Overview evaluates and combines sources into a new, self-contained statement. The court treated that as Google’s own speech.
The search-engine liability shield doesn’t apply
Older rulings protected search engines because they only made third-party content findable. Writing an original answer is a different act.
“You can check the sources” is not a defence
Google argued users can verify for themselves. The court rejected that — the answer is presented as authoritative, and readers aren’t required to cross-examine it.
Important and honest caveat:this is a German, preliminary, non-final ruling and Google is expected to appeal. It is not yet law in the UK or US. But it’s the clearest signal so far of the direction every AI answer engine is heading — and the reasoning isn’t limited to Google. Any platform that synthesises content into a confident answer is exposed to the same argument.
Why this should worry any business, not just publishers
The two businesses in the Munich case only found out because the false claims were severe and visible. Most misrepresentation isn’t. An AI engine can:
Confuse you with another entity
Blend your company with a similarly named one — entity-boundary conflation — and attribute the other’s attributes to you.
Repeat an outdated fact
State old ownership, a closed location, or a former product long after it changed.
Over-weight one negative source
Repeat a single negative source as if it were settled consensus.
Launder a claim
Strip the source, restate it in its own voice, and attach citations that don’t actually support it.
That last one is the dangerous case, because the usual defence (“just click the source”) is exactly what’s been removed. We documented a real instance of it — read the citation laundering case study.
And the damage compounds quietly. By the time a customer mentions that “the AI said something odd about you,” the claim may have been propagating for weeks across multiple engines.
Questions businesses are asking right now
Is Google liable for false information in AI Overviews?
A German court ruled in May 2026 that Google can be directly liable for false statements in its AI Overviews, treating the AI-generated summary as Google’s own content rather than third-party search results. The ruling (LG Munich I, 26 O 869/26) is preliminary, German, and under appeal, so it is not settled law elsewhere — but it is the first clear judicial finding of its kind and its reasoning applies to any AI answer engine.
Can an AI Overview defame my business?
An AI Overview can state false facts about a business that damage its reputation, and a court has now found at least one operator legally responsible for exactly that. Whether it meets the legal threshold for defamation depends on your jurisdiction and the specific claim — but the practical harm (lost trust, lost customers) happens regardless of how a court would classify it.
How do I find out what AI says about my business?
The only reliable way is to observe each AI engine directly and compare its output against a verified record of what’s actually true about your business. Entidex does this continuously across the major AI engines and other surfaces, and flags where what an engine says diverges from the verified record.
What can I do if an AI engine gets my business wrong?
You need dated evidence of what was said, when, and on which surface — the artifact a correction request or legal step depends on. The hard part is that the misstatement is often invisible to you and the provenance is stripped at source, which is why continuous, timestamped monitoring matters more than a one-off check.
How Entidex helps
Entidex resolves your business to a canonical, verified record— the source of truth — and then continuously watches what AI engines and other surfaces actually say about you.
- AI Visibility, Sentiment and Share of Voice tell you whether AI engines see you, how they frame you, and how much of the conversation is yours.
- Trust Stack, Narrative Coverage and Maturity show whether your representation is well-sourced, consistent across engines, and defensible.
- Every divergence between what an engine says and what’s true is captured with a timestamp— standing, dated evidence that survives even when an engine strips the source.
The Munich court noted that verifying these AI claims means comparing the underlying sources against the engine’s output — and that internally, only the operator can do it. Entidex does the equivalent from the outside, for your entity, on a schedule.
Go deeper with Entidex
The full entity intelligence observatory — beyond Explore Entidex
- Continuous multi-source entity observation
- Alerts when sources diverge or drift
- Cross-surface consensus + visibility over time
- Evidence-anchored intelligence reports
This page is commentary on a legal development and general information, not legal advice. The Munich ruling is a non-final German decision under appeal and does not establish law in other jurisdictions.