The AI response about your business is now the platform’s own speech. A German court has now declaredand this changes who is responsible when the answer is wrong. The trial itself is the short story. The most important thing is what a response engine does once it can be held accountable for what it says.
Munich court rules AI preview is Google’s own content
The Munich Regional Court issued a provisional order on May 28, 2026 (case 26 O 869/26) prohibiting Google from repeating false claims its AI Overview had made about two local publishers. The overview linked them to scams and subscription traps, making connections that did not appear in any of the cited sources.
The court treated the AI Overview as Google’s own content rather than a list of search results. In its words, the overview produces “independent, novel, and substantive statements” by evaluating and combining sources, so the liability protections that cover a regular results page do not apply. He rejected Google’s argument that users should check the answer themselves. If the machine writes the sentence, the owner of the machine stands behind it.
Search engines have always turned up bad pages, and the law has long protected them. The court treated the AI insight as being of a different nature. He fabricated a false claim, piecing together fragments from multiple sources into a sentence that none of them contained, and this fabrication is what the court called authorship. It’s the same recombination that makes AI responses useful: the engine takes your page and rewrites it into something newthen presents this as the answer. A court has now looked at the result of this process and characterized it as authorial speech, with liability attached.
The scope here is narrow. This is a regional court, a temporary injunction, decided under European liability doctrine, and a US court working from different rules of speech and intermediaries could land elsewhere. In the United States, the instinct goes in the other direction, toward treating the platform as an immune intermediary. This instinct was built for an era of links and lists, before a machine began to write the sentence itself. It indicates a direction more than it sets one. This direction is consistent with a discovery made a week earlier, according to which being named by an AI doesn’t mean it believes it. Together the two make the shape clear. How an AI response represents your business is both a trust issue and an accountability issue.
Responsibility Makes the Response Engine Cautious
A response engine that can be held accountable for what he says about a company has every interest in covering up, softening or leave out a mark that he cannot verify. This is the side effect of the decision, and it matters more than any case. If the answer is the platform’s own discourse, the rational response is not to suddenly become accurate. It’s about becoming cautious.
The companies she can support, the ones that have a consistent, unambiguous, machine-readable identity on which she can base her claims, become the safe companies to name. Blurries become a risk worth mentioning.
I don’t know if this is happening in such a clear cut, and no platform has announced anything like this. But the incentive only points in one direction. Responsibility makes a system prudent, and a prudent system reveals what it can defend. You can already see the first form of it. Ask an AI about a small or contested company and observe how often it hedges, defers to an official source, or refuses to characterize the company. Responsibility transforms this courtesy reflex into a rule. This transforms machine-readable identity from a citation tactic into something closer to the table stakes. The question stops being “how can I get the AI to name me correctly” and becomes “am I a company that the AI is confident enough to name?” »
An ambiguous company is a risk worth mentioning
Most companies give a machine at least one reason to doubt it. Your name resolves to two or three different legal entities on your homepage, profiles, and past media coverage, and there’s no telling which model is canonical. Your founder’s title says one thing on your About page and another in an interview that the model always trusts. Your product does something specific, but the only place that’s clearly stated is inside an image or PDF that the parser ignores. Your category is obvious to a human reading the page and ambiguous to a machine reading the markupbecause the page never says, in words that a parser can extract, what the thing actually is.
None of this is a content problem as the last decade has taught you to think about content. It’s an identity problem. The model refuses to pretend that it cannot source properly, in the same way that a careful editor writes a sentence that the reporter cannot stand up. This is why content hoarding continues to fail as an AI visibility strategy. Volume does not resolve ambiguity. A company with ten thousand words and three conflicting descriptions is harder to verify than a company whose home page says the same thing true in every direction a machine reads it. The first seems busy to a person and unreliable to a parser. The second seems clear to a person and quotable to a machine.
Check what AI says about you, then correct the facts
You don’t need a lawyer for this. You must be the company whose response engine is safe.
Start with read what the AI already says about you. Manage your brand, products, and category through the engines your customers actually use and read responses like a stranger would. Check the specific things a careful engine will check: is it listing your category correctly, attributing the right products, naming the right people, and avoiding associations that aren’t yours. Do this on all engines, as they will not agree, and the gap between them is your audit. Most companies have never done this once.
Then correct the facts on which the machine is based. Clearly define the entity. Add organizational markup which tells you who you are, what you do and how to confirm it. Keep your identity consistent across the property models read, so the engine never has to choose between two versions of you. This is the identity layer of Machine-centric architecturethe part of the job that makes a business machine readable before it has to please you. The cost of making a mistake has increased with this decision. Not much, because it’s still regional, but it’s not nothing.
So make it a habit, not a one-time audit. Your facts drift, the web around you changes, and patterns recycle. Companies that remain verifiable are those that verify what the response says about them on a schedule, the same way they would verify their own analytics.
Trials will be rare and linked to their jurisdictions. The consequence that counts is slower and structural. When the response involves risk, the engine becomes cautious, and a cautious engine surfaces businesses it can support. Make yours one of them.
More resources:
This article was originally published on No hacks.
Featured image: Viktoriia_M/Shutterstock
