Who Is Responsible For Answers AI Gives You? A German Court Has Some Thoughts | MisalignedSitemapOpen in appSign up<br>Sign in

Medium Logo

Get app<br>Write

Search

Sign up<br>Sign in

Mastodon

Misaligned

Journal and magazine on ethics in AI, technology, data and science.

Press enter or click to view image in full size

Who Is Responsible For Answers AI Gives You? A German Court Has Some Thoughts

A German court has ruled that Google can be responsible for answers given by its AI Overview. That ruling could be significant for AI searches.

Wolfgang Hauptfleisch

6 min read·<br>3 hours ago

Listen

Share

This week, the district court (“Landgericht”) in Munich, Germany, issued a preliminary ruling prohibiting Google to further disseminate untrue factual assertions about two Munich publishers in its “AI Overview”.<br>The AI had incorrectly assigned information about other companies to the plaintiffs. The judgement is significant for the question: Who owns the answers AI is giving you? And what role do “AI summaries” play on the modern web?<br>The background<br>Two Munich based publishers found that when someone asked Google about their companies, Google had something astonishing to say in its “AI Overview”: “Yes, […] is known for dubious business practices and is often perceived as a scam.”<br>The answers were not based on any faulty third party source, but on the fact that the algorithm had wrongly assigned the defamatory statements to the publishers in an associative error, a fact Google did not challenge.<br>It is not the first time this happened, as we explored in “Defamation By Generative AI Is More Than A Glitch”: In June 2023, a Georgia radio host started legal proceedings in the US against OpenAI after the ChatGPT incorrectly stated that he had been sued for “defrauding and embezzling funds” (Loeb & Loeb 2025), and a Georgetown University Professor was falsely accused of sexual harassment by ChatGPT (Utz 2023).<br>What is interested in this case is the finding that the immunity of search engine providers from faulty third party material does not apply.<br>Can Google argue its “AI Overview” is “just another search engine”?<br>In court, Google tried to argue that it is shielded from liability for false statements as it only “summarizes” sources found online, and that most users would understand that AI outputs must be verified independently.<br>It argued that Google is “not itself responsible for data processing, does not adopt the information of third parties as its own in the “overview with AI” and is accordingly only liable if it is made aware of an obvious infringement of rights.” (translated from German).<br>The judges did not agree with this assumption: They ruled that the answers are presented in a form that makes the summaries “Google’s answers”, not just a reference to another source., even if some reference is given that could have cleared up the misunderstanding.<br>If only Google can correct the underlying algorithm and output, then Google owns the answer.

More importantly, there was no third party to blame: It was the amalgamation of the sources and the wrong assignment to the plaintiffs via Google’s algorithm that caused the defamations. The judges explain that while in a traditional search engine the publishers may be able to sue to stop a third party from publishing defamatory statements appearing in Google search results, in case of the “AI Overview” this path is not available, and only Google can correct the underlying algorithm and output.<br>Get Wolfgang Hauptfleisch’s stories in your inbox

Join Medium for free to get updates from this writer.

Subscribe

Subscribe

Remember me for faster sign in

The court went further in dismissing Google’s arguments: They stated that the service’s functionality “would be significantly diminished if the ‘AI overview’ were generally regarded as unreliable”. In other words, if Google would really rely on users to check every reference given in the “AI Overview”, how would the service make any sense or differ from normal search results?<br>“By no means essential for using the internet.”

The judges contend those courts have historically held that search engine providers are not expected to check on every search result, hence shielding search engines from liability, and that search engines, in the way they work, are a necessary part of the internet. The “AI Overview”, however, is not:<br>Furthermore, an “AI-powered overview” — unlike the display of pure search results in search engines — is by no means essential for using the internet.<br>Because even displaying search results via links makes the “flood of data” usable for the individual; the “AI-powered overview,” on the other hand, structures and evaluates data according to a system that is not immediately apparent to the user and thus — depending on the underlying algorithm — also channels the answer to the search query. While many may consider this desirable and a way to facilitate the search process, it is by no means essential for managing the...