A Court Just Made Google Liable for What Its AI Says About You

Something happened in a German courtroom in late May that every business owner should take a quiet minute on. The Regional Court of Munich ruled that Google is directly liable for false statements its AI search summaries made about two local publishers. Not a link Google passed along. Statements the court decided were Google's own words.

The case (26 O 869/26, decided 28 May 2026) turned on Google's AI Overviews, the AI-written answer that now sits above the normal results. For certain searches, that AI answer wrongly tied two Munich publishing companies to scams and dubious business practices, accusations that did not appear in any of the sources it claimed to be summarising. The AI had, in plain terms, made it up.

The court's reasoning is the part worth sitting with. It drew a line between a search engine that lists other people's pages and an AI that writes a fresh, self-contained answer. The first is a messenger. The second, the judges decided, is the author. Google has said the decision is not final and is reviewing it, and German law is not Australian law. But the direction of travel is unmistakable, and the commercial lesson lands long before any appeal does.

Why this matters even if you never see a courtroom

For an Australian small business, the legal fine print is not the real story. The real story is that AI answers now describe you to a customer before that customer ever reaches your website. Someone asks an assistant whether your cafe is any good, whether your trade is reliable, whether your shop is still open, and the AI gives them a confident paragraph. Most of those customers will never tell you what they were told. They just quietly decide.

When the AI gets you right, it works like a free recommendation, the warm word-of-mouth you used to rely on, now happening at machine scale. When it gets you wrong, it does the opposite, and you may never find out why the phone went quiet. The SEO world has been circling this for months. Marie Haynes, who tracks AI search quality and trust closely, has long argued that the signals telling an AI who you are and whether you can be trusted now matter as much as old-fashioned rankings. We made a related point recently in what to do when AI search recommends your competitor instead of you.

The uncomfortable truth is that you do not get a say by default. The AI builds its picture of your business from whatever it can find scattered across the web, and then speaks on your behalf with total confidence, accurate or not. A court in Munich can tell Google off after the fact. It cannot put the customer back who already chose someone else.

What you can actually do about it

You cannot stop AI from describing your business. That ship has sailed, and frankly you would not want to, a good AI answer is one of the cheapest forms of marketing there has ever been. What you can do is make sure the picture the AI is working from is clear, current and unmistakably yours, so the answer it gives leans in your favour rather than against it. This is the heart of generative engine optimisation (GEO): shaping what AI search understands and repeats about you. Here is what good looks like once it is handled properly:

• The AI describes your business accurately, the right services, the right area, the right hours, because the facts it draws on are consistent everywhere it looks.
• When a customer asks an assistant for a recommendation in your category, your name is in the answer rather than a competitor's.
• The trust signals an AI leans on, real reviews, clear ownership, a coherent web presence, all point the same way, so the model treats you as a safe, credible answer.
• You actually know what the major AI assistants are saying about you, instead of finding out from a customer who almost did not come in.
• When something is wrong or out of date, it gets corrected at the source the AI reads, not argued about after the damage is done.

A court can make a platform answer for what its AI said. It cannot give you back the customer who already chose someone else.NextAura

None of this is a one-afternoon job, and it is not a setting you switch on. It is ongoing work: understanding how the models see your business today, fixing what is wrong or thin, and keeping the picture accurate as the AI answers keep shifting underneath you. The Munich ruling is a useful prompt to stop assuming the AI has you right, and to go and check.

This is exactly what we do at NextAura. We work out what AI search is currently saying about your business, make the corrections at the places these models actually read, and build the AI search and GEO foundations that keep the answer in your favour as the technology moves. If you would rather know what AI is telling your customers, and shape it, than find out the hard way, get in touch and we will take it from there.