In a ruling dated September 24, 2026, the Aachen Regional Court ordered Google not to further spread false crime-related statements displayed by its AI search feature, AI Mode, about a local doctor. The court treated AI-generated answers as statements made by Google itself, rather than merely directing users to third-party websites. The ruling is a preliminary injunction and is not yet final.
An unfounded scandal on the other side of the search box
How would you feel if a colleague told you one morning, “You’re showing up on Google”? This case unfolded in Aachen, Germany. A man who was both a professor and head of neuroradiology held the position from 2010 until late June 2026. On August 28, a colleague alerted him that Google’s AI was explaining his departure by linking it to allegations that he had sexually assaulted a secretary.
He had nothing to do with the incident. The AI had confused him with events involving another former university professor. The court confirmed from its own records that the defendant in the criminal case was also someone else.
A fabricated story grows with follow-up searches
The serious problem was that the AI did not stop at a single error. Adding terms such as “suspension,” “dismissal” and “trial” to searches prompted AI Mode to return a succession of detailed, multi-paragraph statements. It even plausibly pieced together a legal outcome that never existed.
With links listed in search results, readers can open the sources and check whether the information is accurate. But an AI answer is presented as a single, polished piece of text. Readers cannot tell when errors have been mixed in.
Weeks passed without action despite request to remove the content
In late August, the doctor explained in detail how the mix-up had occurred and asked Google by email to remove the content. Google reportedly replied in mid-September that it would take no action. The doctor’s formal legal demand, sent the same day, was confirmed as received, but the court found that the false statements were still being displayed.
The Aachen Regional Court prohibited the dissemination of three types of disputed statements. It found that the false information could seriously damage the doctor’s professional reputation and issued the emergency measure without hearing Google’s position. Violations can incur a fine of up to €250,000 each. That amounts to tens of millions of yen, and repeated violations could carry a substantial cost.
A guide to links—or the author of the text?
The most noteworthy aspect of this ruling is how it viewed Google’s role. The court said, in effect, that Google is responsible for answers generated in the course of providing AI-powered search. It also cited language from an earlier Munich ruling, which said that AI evaluates and combines content from multiple third-party websites to create independent, new substantive content.
Traditional search engines were more like library guides. They could tell you which shelf held a book, but responsibility for the book’s contents rested with its author or publisher. AI search, by contrast, is like a guide who reads the books on the shelves and starts summarizing them in its own words. If what it says is wrong, the guide itself can be held responsible. Seen that way, the court’s reasoning makes intuitive sense.
The court also pointed to steps Google could have taken. For example, it could have checked AI output against the source websites it referenced. The court also said that confronting the AI with the doctor’s objection should have allowed it to recognize the error. The reasoning is that if Google is to be held responsible, there must be ways to prevent the harm.
What the Munich settlement left unresolved
The Aachen ruling did not come out of nowhere. In late May 2026, the Munich I Regional Court found Google directly responsible in a case in which its AI Overviews had falsely linked two local publishers to a fraud scheme. The court treated AI Overviews not as ordinary search results, but as statements made by Google itself.
Google publicly said it intended to appeal, but the case ended in a settlement before an appellate court could rule. As a result, no higher court had the opportunity to review the Munich court’s reasoning. The Aachen ruling is also not yet final, and Google can challenge it. It is too early to say that a legal standard has been established across Germany, but it remains a fact that another court followed the same line of reasoning.
Could the effects also reach those who provide material for the answers?
If AI answers count as content produced by Google itself, the issue extends beyond how to address misinformation. In theory, writers of web articles used as source material for answers could also have grounds to assert their own rights. In the most extreme scenario, AI providers could have to pay source creators each time they generate an answer. These issues were not contested in this case; they are possible developments in the future.
Another concern is the pattern of individual settlements. Unless a case reaches the highest court, no binding precedent is established. Specialized law firms or litigants could repeatedly pose targeted questions, get AI to piece together inaccurate text, then seek preliminary injunctions and use them to force settlement talks. Some warn that this could become more widespread. Because AI search is easy to steer, responsibility may need to be clarified sooner rather than later.
As search shifts from links to prose, the question of “who is speaking?” carries increasing weight. So far, the German courts’ answer has been: “Google is speaking.”
Comments