Google
is liable for
AI mode.
is liable for
AI mode.
of
Who is liable if an AI search attributes one person’s criminal offense to another? And is it sufficient for the search engine operator to point out that its technology is prone to errors? The Aachen Regional Court has issued a preliminary injunction prohibiting Google from providing incorrect answers in AI mode.
The AI Confuses Two Professors
The petitioner is a physician. He served as chief physician and professor at a university hospital until the end of June 2026 and has since left that position. On August 28, 2026, a fellow physician brought a Google search to his attention. Anyone who entered his name along with the word “leave of absence” received a detailed response in AI mode.
AI Mode is a feature of Google Search in which an AI generates its own responses. The response attributed a leave of absence, a criminal conviction, and several labor court cases to the doctor. None of these things had anything to do with him. The incidents referred to another former university professor.
That same day, the doctor sent a written request and an email to Google demanding the removal of the content and describing the facts of the case. Google responded that it would not take any action. Even a formal cease and desist letter from a lawyer dated September 15, 2026, was to no avail. When the doctor conducted another search using additional keywords, similar texts appeared. The doctor then filed a motion for a preliminary injunction—that is, a court order issued in summary proceedings.
The Aachen Regional Court prohibited the statements
Inits decision dated September 24, 2026—Case No. 8 O 268/26— , the Aachen Regional Court prohibited Google from using the three contested text versions in AI mode.
Are AI responses even statements of fact?
In the court’s view, the texts constitute statements of fact. Whether someone was dismissed, convicted, or fired is a matter that can be proven. The statements constituted an infringement on the doctor’s general right of personality—that is, his right to respect for his person and his reputation.
The doctor had substantiated the falsehood through an affidavit. According to its own statement, the chamber was also aware that the criminal proceedings in question had in fact been conducted against another person.
Why is Google liable for the text generated by its AI?
The central question was whether Google is liable as a “contributor to the infringement.” A “contributor to the infringement” is anyone who causes an infringement themselves or knowingly contributes to it, regardless of fault. In the case of traditional search engines, the case law of the Federal Court of Justice distinguishes between whether the operator merely makes third-party content discoverable or adopts it as its own. Accordingly, for mere search results, a duty to review arises only when the operator is specifically notified of a clear legal violation.
The Aachen Regional Court takes a different view of AI mode. It holds that Google provides the AI-powered search and is therefore responsible for the answers generated by it. The court explicitly bases its ruling on the decision of the Munich I Regional Court. According to that ruling, an AI search does not merely display third-party websites but makes independent, new statements based on the analysis and linking of third-party content. The operator can verify such statements, at least by cross-checking them against the underlying sources. We have already discussed the Munich I Regional Court case in detail.
The respondent could easily have verified this herself. Had she run the applicant’s objections through her own AI, she would have concluded that the statements were false.
Added to this was the fact that the doctor had brought the error to Google’s attention twice—first with his request for removal, then with the cease and desist letter. Google had actively refused to take action in accordance with the court’s findings.
How to Respond to Incorrect AI Responses
If an AI search spreads false information about you or your company, there are five things you should keep in mind.
- Document the responses. Save screenshots with the search term, date, and time. AI responses vary from query to query, and it is often no longer possible to verify them later.
- Provide specific information to the operator. Describe which statement is false and what the correct information is. In the Aachen case, the court also based its ruling on the fact that Google had comprehensive knowledge of the facts.
- Make the falsehood verifiable. Have documentation ready that shows the true facts of the matter. In summary proceedings, an affidavit may suffice.
- Act quickly. In the Aachen case, less than four weeks elapsed between the colleague’s initial report and the decision. If you wait too long, you risk a court denying that the matter is urgent.
- Clarify your legal strategy. Whether a cease-and-desist order, a request for withdrawal, or taking action against the source sites makes more sense depends on the individual case. Consulting with an expert on artificial intelligence can help you assess the situation.
For companies that provide AI-powered responses themselves—such as through a chatbot or search function—the opposite applies. Under this ruling, it is not sufficient to simply dismiss complaints about false statements regarding individuals. It is advisable to have a functioning review and removal process in place.
Conclusion
The outcome comes as no surprise. The Aachen Regional Court follows the Munich I Regional Court and applies its reasoning to AI mode. This is favorable for those affected, as AI searches are not treated as mere lists of results. For Google and other providers of artificial intelligence under IT law, it means they must take responsibility for any errors made by their systems. We have previously reported on a similar ruling regarding the AI bot Grok on the X platform.
We’d be happy to
advise you on
AI!
