Google's AI Overview Can Now Lie About Your Business. A German Court Just Said That's Google's Problem, Not Yours.

A client called me last month, voice tight, and read me a sentence off his screen: his company was "known for dubious business practices." He didn't write that. No journalist wrote that. No review said that. Google's AI Overview generated it, sitting at the very top of the page, above every real result, in that calm confident gray box people now read instead of clicking anything.

He runs a 14-year-old B2B firm with a clean record. And the first thing a prospect saw when they Googled him was a fabricated accusation presented as fact.

Here's what changed in June 2026: that fabricated accusation is now, in at least one courtroom, legally Google's fault. Not the websites it pulled from. Google's.

The AI Overview Said My Client Was "Known For Dubious Business Practices"

The maddening part wasn't the lie. It was the powerlessness. He clicked the sources Google cited under the summary, and none of them said what the summary said. The AI had stitched together a few unrelated mentions, added an evaluative judgment nobody made, and served it as a neutral answer.

That's not an edge case. A study reported in April 2026 estimated AI Overviews generate somewhere between hundreds of thousands and millions of false statements per hour. Per hour. When you search your own brand and find a confident summary that's subtly or wildly wrong, you are not unlucky. You are inside the normal failure rate of the system.

For a long time the advice was basically "good luck." You could thumbs-down the box into the void. You could try to fix the underlying pages and hope. The question of whether Google itself owed you anything was, legally, untested. If you want the broader strategic version of this, I wrote a full companion piece on how to control what AI search says about your brand. But the legal ground just shifted, and that matters.

What the Munich Court Actually Ruled (And Why It's a Big Deal)

On roughly May 28, 2026, the Regional Court of Munich (Landgericht München, Case No. 26 O 869/26) issued a preliminary ruling that two Munich publishing companies had been defamed by Google's AI Overviews. The summaries had falsely tied them to scams, subscription traps, and "dubious business practices," inventing connections that did not exist in the cited sources.

The court's core finding is the part that should make every SEO and business owner sit up: AI Overviews are Google's own content, not neutral search results. The judges held that Google bears direct liability because Google alone controls the AI, the model, and the algorithms that produce the text. Google was ordered to cover roughly 80% of the legal costs. This is the heart of the emerging google ai overview liability question, and a court answered it bluntly: if your machine writes it, you own it.

Two honest caveats so I'm not overselling. It's a temporary injunction, not a final judgment, and Google can appeal. And it's German. But precedent like this doesn't stay in one country quietly.

Why "It's Just Pulling From the Web" Stopped Being a Defense

Google made two arguments in Munich, and the court rejected both. That's the precedent-setting bit, so let me be specific.

Argument one: we're a neutral intermediary just surfacing what the web says. The court's answer was that AI Overviews "rewrite, combine, and evaluate information in its own words," producing "independent, new, and substantive statements." Translation: the moment the system stops quoting and starts paraphrasing-plus-judging, it's authoring. My client's case was exactly this. No single source called him dubious. The AI did, by combining fragments and adding a verdict.

Argument two: users can verify for themselves. The court said the mere possibility of disproving a statement through further research does not exempt Google from liability. You don't get to defame someone and then point out they could theoretically prove you wrong.

If you've ever had a boss or a client wave a screenshot of an AI answer at you, you already know how fast a machine-generated claim becomes "true" in someone's head. I wrote about that exact panic in what to do when your boss screenshots a ChatGPT answer recommending your competitor, and the psychology is identical here: confident formatting beats accuracy every time.

Germany vs the US: Why This Doesn't (Yet) Save American Businesses

Most of my clients are in the US, so I'll be straight with you: you probably cannot walk into a German court next week. German press and personality law is genuinely stronger and more plaintiff-friendly than what you've got in the States, where Section 230 has historically shielded platforms for third-party content.

But "historically" is doing a lot of work in that sentence. Section 230 was written for hosting other people's words. It is genuinely unsettled whether it covers content the platform's own AI generated, and nobody has gotten a clean answer yet. Meanwhile the pressure is already building stateside: Chegg is suing Google over AI Overviews, and a Rolling Stone publisher lawsuit was ongoing as of January 2026 (Reuters). The Munich ruling won't bind a US judge, but it hands every American plaintiff's lawyer a well-reasoned roadmap for arguing google ai overview liability. Precedents travel. This one has a passport.

How to Actually See What AI Overviews Say About Your Brand

Before you can fix the lie, you have to find it, and most business owners have genuinely never looked. So here's the monitoring routine I run for clients.

Search your brand the way a suspicious prospect would, not the way a fan would. Type "is [brand] legit," "[brand] scam," "[brand] complaints," "[brand] reviews." Those query patterns are where the damaging summaries surface, because that's the intent the AI is trying to satisfy. Run them logged out and in an incognito window so you see what a stranger sees, not your personalized bubble. Set up brand alerts so you're not checking manually forever. And every time a summary appears, read the cited sources underneath it and compare. The gap between what the sources say and what the summary claims is your evidence, and your lever.

The Removal Playbook: Feedback, Sources, and the Opt-Out Toggle

Here's the order of operations I use, weakest-but-fastest to strongest-but-slowest. None of these is magic alone. Stacked, they work.

Step one: the feedback form. Thumbs-down the Overview and flag it through Google's feedback tool. It's fast and it costs nothing, but on its own it rarely fixes anything. Do it anyway, because it creates a record and it's the only direct signal you've got.

Step two: fix the source data. This is the real work and the part that actually moves the needle. The AI pulls from your Google Business Profile, your schema markup, third-party listings, and the contact pages of whatever sites it cites. Clean those up. Correct the wrong listings. Tighten your structured data so the machine has accurate raw material. Google's track record on moderating its own data is uneven, by the way: they deleted 292 million reviews in a single year, which tells you both that they're trying and that a lot of garbage gets through.

Step three: the opt-out toggle. As of June 3, 2026, Google rolled out a Search Console setting to opt your site's content out of AI Overviews, with enforcement starting June 17, 2026. Two things to know: it excludes the Gemini app, and Google says it carries no organic ranking penalty. If you want to understand what AI is doing to your traffic before you flip it, I broke down what the Search Console AI traffic report actually shows. Opting out is a real option now, but it's a tradeoff, not a free win: you lose the visibility too.

What This Means for SEO and Reputation Going Forward

For fifteen years, ranking was about traffic. AI Overviews turned the top of the search page into a reputation surface, and that's a different job. The summary is now the first impression, and you don't write it, a model does, and that model can be wrong about you to millions of people at once.

So the work splits. You still want to rank, and the rules for showing up in AI answers are more knowable than most people pretending to be GEO experts admit, which is the whole point of my guide to ranking in AI search. But alongside ranking, you now need a defensive layer: monitoring what's said, fixing the sources, and knowing your removal options. The Munich ruling matters here beyond the courtroom, because it reframes the conversation. The summary isn't an act of God you endure. It's content somebody is responsible for, and the question of google ai overview liability increasingly answers that the somebody is Google.

The Honest Bottom Line

I won't tie a bow on this, because there isn't one yet. You probably can't sue Google in Munich, the toggle is a tradeoff, and the feedback form is mostly shouting into a box.

But three things are true at once. The false summaries are real and they're hitting clean businesses at scale. The legal ground under Google just cracked, in public, with a reasoned ruling other courts can borrow. And the leverage you actually have today, monitoring and source-fixing, is boring, unglamorous, and effective.

My client's Overview is fixed now. It took a feedback flag, a week of cleaning up listings and schema, and patience. Not a lawsuit. If you haven't searched your own brand the suspicious way in a while, do it this afternoon. And if what you find makes your stomach drop, that's the part Wunderlandmedia can actually help with.