
use as a
trademark
by AI.
use as a
trademark
by AI.
from
Does a search engine commit an infringement on trademark rights if its AI feature mentions the name of a well-known perfume brand when searching for similar scents? Is it sufficient for trademark use if an AI automatically summarizes third-party content and uses trademarks in the process?
Trademarks in AI Summary
Anyone searching for a brand-name perfume on a major search engine will now often see not only the usual search results but also an AI-generated summary right on the results page. It was precisely these summaries that were the subject of summary proceedings before the Berlin Regional Court.
A perfume manufacturer sought a court order to prevent Google’s AI features from mentioning its brand names in connection with inexpensive imitations—so-called “fragrance twins” or “dupes.”
At the heart of the decision lies a question that is central to trademark law as a whole: Does the mention of a trademark by an AI search function constitute use of the trademark?
What had happened?
Google offers two AI features: an automatically generated “AI Overview” that appears above the search results, and an “AI Mode” where users can ask questions directly. Both features summarize content from third-party sites and provide links to them.
In response to the request “Name some fragrance dupes” or a search for fragrance dupes of a specific brand-name perfume, the AI generated a list of well-known sellers of fragrance dupes and provided links to their websites. For some queries, ads were also displayed under the heading “Sponsored Products,” which also featured dupes.
The perfume manufacturer viewed this as an infringement of its EU trademarks and a violation of competition law. It sought a preliminary injunction prohibiting the search engine from mentioning its trademarks in this context in the AI-generated texts.
Decision by the court
The Berlin Regional Court II dismissed the motion for a preliminary injunction in its judgment dated June 1, 2026 – 52 O 62/26 eV.
Is there trademark use?
Only those who “use” a trademark in the legal sense can be subject to an injunction. Merely mentioning a trademark name is not sufficient. The decisive factor is whether a company actively uses the mark in its own commercial communications.
In this case, the court did not find that the mark had been used within the meaning of trademark law.
According to the case law of the European Court of Justice, use as a trademark requires that, from the perspective of third parties, a sign appears to be an integral part of a company’s own communications and serves to promote its goods or services. The AI-generated texts do not meet this requirement, according to the Berlin judges. They merely summarize content from third-party websites without the search engine giving the impression that it is its own advertising or its own statements.
No appropriation of third-party content
Trademark use could also be established if the search engine adopts the third-party content as its own, i.e., assumes responsibility for it outwardly. The court rejects this argument as well. A reasonably informed and reasonably observant user can easily recognize, based on the numerous links, snippets, and preview images, that this is an automated summary of third-party sources and not editorial content produced by the search engine itself. An explicit disavowal by the search engine of the AI-generated texts is not required for this.
No decisive influence on the content
For the use to constitute trademark use, the search engine would also have to be able to control the content of the search results itself. In the court’s view, this requirement is also not met. Which sources the AI draws upon for its response depends solely on which information published by third parties matches the respective user query according to the patterns and structures recognized by the AI. The search engine does not exert any targeted influence over this—unlike the operator of an on-site search engine, who controls the selection of the results list themselves and uses trademarks as keywords as part of their own advertising .
Advertisements do not constitute use as a trademark either
According to the court, even the ads displayed under “Sponsored Products” do not constitute trademark use by the search engine. According to the case law of the European Court of Justice, trademark use cannot be based solely on the fact that the search engine operator receives compensation from its customers for displaying advertisements. Even if these ads appear in the context of AI-generated results, the search engine merely provides the technical conditions for the discoverability of certain websites—it does not thereby associate itself with the advertised products.
A reasonably informed and reasonably observant user would recognize that, through its AI features, the respondent has merely created a new format for search results by having the AI summarize search results based on calculations, without the respondent needing to distance itself from them.
No competitive relationship between the perfume manufacturer and the search engine
Claims under competition law also failed. For that to be the case, the parties would have had to be competitors, meaning they would have had to be in a concrete competitive relationship with one another.
However, a perfume manufacturer and a search engine offer fundamentally different services. The court also found no economic interdependence. The advantage the search engine derives from more attractive AI features has no competitive bearing on the manufacturer’s perfume products.
A so-called indirect competitive relationship might also have been conceivable. This would have been the case if the search engine had specifically promoted sales by the counterfeit sellers who compete directly with the perfume manufacturer. However, the court rejected this argument as well. The mention of alternative sellers in the AI-generated texts promotes their sales, at most, “instinctively” as a side effect, without this constituting a targeted commercial act on the part of the search engine. Furthermore, the links to third-party sellers in the AI-generated texts are unpaid, meaning the search engine derives no direct economic benefit from them.
Conclusion
The decision by the Berlin Regional Court II stands in contrast to rulings such as that of the Munich Regional Court I regarding Google’s AI-generated summaries, and the reasoning of the Frankfurt Regional Court also points to a different outcome. In those decisions, the content was attributed to Google precisely because of the AI’s independent linking and structuring capabilities.
The Berlin Regional Court II describes the search engine’s AI functions in almost identical technical and functional terms, namely as the summarization, structuring, and formatting of third-party content into a newly generated text formulated in its own words. Nevertheless, it reaches the opposite conclusion and denies that the search engine has appropriated the content.
This difference can hardly be explained solely by the distinction between trademark law and personality rights or competition law, since the same technical characterization of the AI output leads to attribution as an independent statement in one case but not in the other. Given these parallels, the decision could well have gone the other way.
We are happy to
advise you about
AI!







