No more neutral search engines: Google and Perplexity in the crosshairs – Why AI search engines are suddenly liable as publishers

Copyright vs. AI: Why press publishers are now drawing new hope in the fight against Google

The internet is facing a historic turning point: When answers are generated by artificial intelligence, search engines are suddenly considered publishers – with far-reaching legal and economic consequences. The German media regulator (ZAK) has now initiated precisely this paradigm shift with unprecedented rulings against Google and Perplexity. Where previously the lucrative liability privilege applied to neutral platforms, full journalistic responsibility now comes into play. This is because AI search engines and chatbots no longer simply forward information; they link, modify, and create entirely their own content.

For press publishers and media professionals, this development represents a much-needed glimmer of hope. While AI-generated answers offer users unparalleled convenience, they also deprive the actual creators of information of significant traffic, revenue, and visibility—an existential threat to journalism. The media authorities' decision thus marks far more than a mere legal footnote: it attacks the economic foundations of tech giants, renders established safeguards like the Digital Services Act partially inapplicable, and raises the pressing question of who will ultimately control the interpretation of information in the digital age. Below, you will find a detailed analysis of the new legal situation, its underlying economic structures, and the potential consequences for the World Wide Web.

When the answer itself becomes the news: How Google and Perplexity suddenly become publishers – and why this is reshaping the power dynamics on the internet

A regulatory first strike with a signaling effect

The German media regulator has set a precedent that touches upon the foundations of digital information dissemination. The Commission for Licensing and Supervision (ZAK) has, for the first time, issued rulings against AI services from Google and Perplexity, thereby explicitly applying German media law to AI search engines and AI chatbots. The ZAK is the central supervisory body, comprised of the heads of the fourteen state media authorities, and is responsible for the licensing and oversight of nationwide private broadcasters and online media. Specifically affected are Google's AI-generated search overviews, the so-called AI Overviews, and the AI ​​chatbot Perplexity, along with its AI news page. The underlying proceedings, conducted by the Hamburg-Schleswig-Holstein Media Authority and the Berlin-Brandenburg Media Authority, centered on the question of whether AI-generated responses merely relay third-party content or whether they themselves become independent journalistic products. The regulators' answer is unequivocal: AI responses are not neutral information dissemination but rather the respective providers' own content.

ZAK Chairman Thorsten Schmiege formulated the core message unequivocally, declaring that AI search engines and chatbots are content providers and that German media law will henceforth be consistently applied to them. This assertion is based on a legal opinion commissioned by the media authorities and prepared by legal scholars Jan Oster and Christoph Busch, which systematically examines the media law and regulatory classification of the integration of AI applications into search engines. The decision marks a break in that it departs from the previous debate on training, copyright, and source citations, instead making the generated response itself the subject of legal evaluation.

Related to this:

• If AI lies, Google is liable! The Munich ruling against Google's next-generation search engine

Why the liability privilege of the Digital Services Act fails here

The central legal argument of the media authorities concerns the applicability of European platform regulation. The liability privilege enshrined in the Digital Services Act, according to which platform operators are generally not liable for unlawful content provided by their users, does not apply to AI-generated responses, according to the ZAK (Commission for Licensing and Supervision), because these are not user-provided information but rather content generated by the provider itself. The legal opinion by Oster and Busch supports this position with a nuanced dogmatic justification: The liability privileges of Articles 4 to 6 of the Digital Services Act, by their very wording, refer to information provided by a user, and AI responses do not structurally meet this criterion.

This classification becomes particularly clear in the case of so-called hallucinations, i.e., content invented or incorrectly compiled by AI, which is undoubtedly to be considered the provider's own content. However, even in the case of the mere processing of existing third-party sources, the experts argue for the output to be considered independent, because the repackaging, condensing, and blending of content constitutes a separate editorial act. A different conclusion should only be reached if it remains clearly recognizable to users that only third-party content is being reproduced unchanged. In practice, this exceptional case is rare, since generative AI systems, by definition, rephrase, shorten, and synthesize content from multiple sources.

A Munich court is moving in the same direction

The media authorities' decision is not an isolated case, but rather part of a series of rulings that follow the same legal logic. Just a few weeks earlier, the Munich I Regional Court had ruled in a landmark decision that Google is directly liable as an infringer for false claims in its AI Overviews. The court explicitly classified the AI-generated overviews not as neutral search results lists, but as independent, summarizing statements by the corporation. The court reasoned that the AI ​​function independently summarizes and structures third-party content and condenses it into its own statement, thus making the presentation significantly more comprehensive than traditional search results.

In this case, Google could not invoke the established liability privileges for pure search engine operators, who are typically only liable to a limited extent and only after specific notification regarding linked third-party content. The attempt to apply the case law on autocomplete functions to AI-generated overviews also failed before the Munich Regional Court. Furthermore, the court found that Article 6 of the Digital Services Act did not preclude national injunctions. The court's reasoning is noteworthy: the AI-generated overview is by no means essential for using the internet, but merely an additional service for which Google, as the operator of its own system, is fully responsible. Two completely independent proceedings, a civil court and an administrative authority, thus reached the same legal conclusion: whoever generates answers also bears the responsibility for them.

The logic of media intermediaries and the prohibition of discrimination

Besides the question of liability, a second, economically at least equally significant aspect comes to the fore: the discoverability of journalistic content. When using AI Overviews, the AI-generated answers are prominently and clearly displayed as an essential component of the search results. This makes the traditional list of external links less visible compared to the AI-generated answers, which media authorities consider unlawful discrimination. The core of the accusation is that Google's AI Overviews appear so dominant above the actual search results that traditional links, especially to journalistic sources, are effectively pushed into the background.

The chatbot Perplexity is also the target of a related, but independent, line of reasoning. When an AI chatbot appends third-party content as sources, further information, or entire lists of links to its generated answers, it also significantly influences the discoverability of this third-party content. According to the media authorities, this functionality fulfills the criteria of a so-called media intermediary, i.e., a service that aggregates, selects, and presents journalistic and editorial content from third parties. Since the Interstate Media Treaty came into force in November 2020, media intermediaries have been subject to specific transparency and non-discrimination obligations under German law. Providers must disclose the central criteria of their aggregation and sorting logic in easily understandable language and may not systematically disadvantage journalistic and editorial content without objective justification. Therefore, Perplexity and Google will have to be held accountable for these diversity assurance obligations not only as content providers but also as intermediaries.

The economic deep structure: traffic, refinancing, and bargaining power

Behind the legal facade lies a stark economic problem, described in the legal opinion by Oster and Busch as a substitution process. The integration of generative AI into search engines is structurally changing how information is sought online, because a self-contained, flow-text-based answer is increasingly replacing a linking results list. This substitution process reduces the traffic flowing from the search engine to the original journalistic sources, thus directly impacting the financing of high-quality, research-intensive content in particular. For publishers, every user who consumes an answer directly in the AI ​​overview without clicking on the original source represents lost advertising revenue, a lost potential subscriber, and a lost data point for their own audience measurement.

From this substitution effect, the experts derive two interconnected risks: a threat to the diversity of content and opinions, and a shift in bargaining power in favor of AI search engine operators. Whoever controls the interface of visibility effectively also controls which journalistic offerings can survive economically and which cannot. This power shift is not an abstract fear, but a direct consequence of the technical architecture of AI search systems, which are designed to provide users with a comprehensive answer without requiring any further clicks. This very user-friendliness, which represents progress from the perspective of an individual search, becomes a structural problem from the perspective of the entire journalistic ecosystem.

The industry association Corint Media, which represents the copyright and related rights of broadcasters and press publishers, welcomed the ZAK rulings as an important regulatory signal. Managing Director Christine Jury-Fischer explained that AI-powered services offered by global platforms do not operate outside the bounds of current media law, and that anyone using AI to suppress journalistic content and determine its discoverability must be held accountable for transparency, non-discrimination, and ensuring diversity. If the visibility and financing of journalistic content are not guaranteed, media diversity itself will ultimately come under pressure. This assessment from a directly affected economic actor aligns perfectly with the macroeconomic analysis presented in the legal opinion.

Xpert.Digital is a data-driven B2B industry hub led by Konrad Wolfenstein . The company acts as an external, quasi-in-house solution for industrial partners, closing operational gaps in marketing, content, and sales – without requiring additional resources on the client side.

More information here:

• The quasi-in-house solution: How Xpert.Digital closes operational gaps in B2B marketing and sales – Smart Content-Driven Business

Between fundamental rights protection and regulatory mandate

The regulation of AI search engines does not take place in a legal vacuum, but rather in a finely balanced field of tension between fundamental and human rights. AI systems themselves are not entitled to fundamental rights, but the operators of these systems, as companies, enjoy the protection of freedom of occupation and entrepreneurial freedom, as well as, in principle, freedom of communication. Therefore, any regulatory intervention must be measured against strict proportionality standards. The expert opinion cites the following criteria as guiding principles for this balancing of interests: the human accountability and autonomy of AI expenditures, protection against the distortion of individual and public opinion, the protection of personal rights, and equal communicative opportunities between different providers.

This consideration explains why media authorities do not categorically prohibit or sanction every AI response, but instead regulate it selectively at two points: responsibility for one's own content and non-discrimination in the sorting of external content. The legal mandate of the media authorities, which ZAK Chairman Schmiege explicitly emphasized, is to safeguard the diversity of journalistic and editorial media, which would otherwise be threatened with disappearing behind opaque algorithmic ranking decisions. This mandate is enshrined in the Interstate Broadcasting Treaty and has so far primarily focused on traditional search engines, social networks, and news aggregators. The ZAK rulings now explicitly extend its scope to generative AI systems for the first time.

Related to this:

• From search box to answer engine: The brutal "winner-takes-all" battle for AI truth

The unresolved side question: Are AI search engines even search engines in the sense of the DSA?

One of the most complex dogmatic questions concerns the classification of AI search systems within the categories of the Digital Services Act itself. Insofar as AI systems directly access web-based information to answer user queries and simultaneously display traditional search results including links, they can functionally be classified as online search engines within the meaning of Article 3(j) of the Digital Services Act. However, the relationship between this specific search engine concept and the three general intermediary service categories of Article 3(g) is not systematically and coherently resolved. The most convincing solution proposed by the opinion is to understand online search engines as a kind of intermediary service sui generis, which follows its own rules.

Even if one accepts this classification, a key legal uncertainty remains. Because AI responses are classified as original content, a regulatory gap arises between the liability exemptions and the procedural due diligence obligations stipulated by the Digital Services Act for mere intermediary services. The law was designed to regulate the forwarding of external content originating from third parties, not to address the creation of new, independent content by the platform itself. This gap is the primary reason why national media regulators and civil courts in Germany are now beginning to fill the resulting void using the instruments of traditional media and freedom of expression law.

Copyright and the question of fair remuneration

Parallel to the media law debate, the copyright dimension of the problem is intensifying. For the training of AI models, the text and data mining regime under Section 44b of the Copyright Act and Article 4 of the DSM Directive is central in Europe. However, its scope and the practical effectiveness of the machine-readable opt-out, which allows rights holders to exclude their content from AI training, are unclear or inadequately implemented in practice. For the specific use of journalistic content, the press publishers' ancillary copyright under Sections 87f et seq. of the Copyright Act also comes into play.

The report's key economic observation is that AI-generated search results can have a substitutive effect, directly impacting the investment of press publishers. A publisher invests considerable resources in research, fact-checking, and editorial preparation, but if an AI-generated result reproduces the outcome of this investment in a condensed form free of charge, without the user ever accessing the original source, the economic basis of this investment is undermined. In practice, individual publishers' attempts to enforce their rights face significant limitations in terms of proof and enforcement, because it is virtually impossible to trace which specific source, and to what extent, influenced a particular AI-generated result. Therefore, strengthening collective rights management through collecting societies and shifting the burden of investigation and transparency obligations to the operators of the AI ​​search engines themselves, who are technically best positioned to disclose the origin of their generated content, seems a logical step.

The role of the AI ​​regulation and its gaps

The European AI Regulation addresses the topic of AI search engines through several points of contact, without yet providing a fully coherent regulatory architecture. Article 50 obliges providers to comply with transparency requirements regarding synthetic content, while the regime for general-purpose models in Article 51 and subsequent sections, in particular Article 53(1)(c) and (d), stipulates further obligations. Providers of such basic models must, among other things, maintain a copyright compliance policy and publish a sufficiently detailed summary of the content used for training.

The report particularly criticizes Article 50, paragraph 4, subparagraph 2 of the AI ​​Regulation, which concerns synthetic texts on matters of public interest, as being technically unclear. This provision, at least insofar as it relates to AI search engines, should be clarified through codes of practice or guidelines, or, if necessary, strengthened through legislation. At the national level, there is also a need for coordination between the technical market supervision provided for in the German implementing legislation and effective media law enforcement powers that must also cover non-journalistic AI search engine providers. This dual fragmentation of oversight—between technical AI regulation at the European level and safeguarding diversity at the state level—significantly hinders coherent enforcement.

The Digital Markets Act as a complementary competitive instrument

Beyond the media law and copyright law dimensions, the debate also has a tangible competition law component. When a corporation like Google places its own AI response more prominently than competing services or external journalistic sources, the accusation of self-preferential treatment arises—a pattern of behavior that the Digital Markets Act explicitly prohibits for so-called gatekeeper platforms. Furthermore, discriminatory discoverability practices and the gradual consolidation of conglomerate platform power become relevant under competition law when a single corporation simultaneously acts as a search engine, AI response system, and advertising marketer.

While the Digital Markets Act remains a key enforcement framework, it cannot fully replace the specific diversity-safeguarding objective of media law because it is primarily geared towards fair competition between companies and not towards protecting democratic opinion-forming as such. This dual structure, in which competition law protects economic fairness and media law protects democratic diversity, explains why media authorities had to take action in parallel with antitrust proceedings and could not rely solely on the Federal Cartel Office.

What the experts' recommendations mean for the future

From their comprehensive analysis, Oster and Busch derive a multi-stage agenda for adaptation that goes far beyond the current ZAK rulings. The Digital Services Act must first be amended to clarify whether and to what extent AI search engines should be covered at all, and what liability and due diligence framework should apply to the answers they generate. The AI ​​Regulation and its implementing instruments must define the currently vague transparency obligations for AI search engines more precisely so that users can actually understand how an answer was generated.

Copyright law requires further development of the text and data mining opt-out mechanism, supplemented by practical remuneration and enforcement mechanisms that do not overburden smaller publishers. The fourth recommendation goes the furthest: To ensure the systematic classification of hybrid services such as AI search engines between the regime for intermediaries and the regime for content providers, a separate telemedia category specifically for AI search engines should be created at the state level. This new category would need to include, among other things, clear responsibility for content, binding obligations for source attribution and linking, transparency regarding the selection and ranking parameters used, an explicit prohibition of discrimination against journalistic content, a designated agent for service of process in Germany, specific discoverability requirements for particularly trustworthy content providers, and its own transparency rules for algorithmic targeting.

Consequences for companies, publishers and the digital public

For companies affected by AI-generated false statements, the new legal situation opens a clear path to claims for injunctive relief and correction under general freedom of expression law, because AI responses are no longer considered privileged, neutral search results. For publishers and journalistic media companies, recognition as media intermediaries means that Google and Perplexity will have to transparently disclose the criteria they use to cite, link to, or omit journalistic sources in their AI responses. For operators of AI search systems themselves, the new regulatory practice represents a significantly increased legal risk, which in the medium term could lead to more cautious response formats that rely more heavily on source citations in order to minimize liability risks and avoid regulatory sanctions.

For the digital public as a whole, the question arises whether a growing proportion of information dissemination is being handled by a few centralized AI systems whose internal weighting logic remains largely opaque. The affected providers can appeal the ZAK rulings, meaning that judicial review and potentially further appeals are to be expected. Regardless of the specific outcome of these legal proceedings, however, the German media regulator has already clearly answered a fundamental question with its actions: those who generate answers on a large scale that are consumed by millions of people can no longer hide behind the fiction of merely being a neutral intermediary of third-party information. The search engine of tomorrow bears the journalistic responsibility of a publisher, even if it continues to outwardly appear as a purely technical tool.

☑️ Our business language is English or German

☑️ NEW: Correspondence in your native language!

 

I and my team are happy to be available to you as your personal advisor.

You can contact me by filling out the contact form here wolfenstein@xpert.digital:or simply call me at +49 7348 4088 965. My email address is

I'm looking forward to our joint project.

 

 

☑️ SME support in strategy, consulting, planning and implementation

☑️ Creation or realignment of the digital strategy and digitization

☑️ Expansion and optimization of international sales processes

☑️ Global & Digital B2B trading platforms

☑️ Pioneer Business Development / Marketing / PR / Trade Fairs

In industrial B2B, sustainable business relationships rarely emerge overnight. They develop step by step – through visibility, professional relevance, recurring touchpoints, and growing trust. Xpert.Digital's 4-stage model addresses precisely this: It offers a structured path that begins with a manageable entry point and can evolve into deeper collaboration in business development if needed.

Instead of relying on loud marketing promises, this model puts the relationship at the forefront. Companies start with clearly defined, easily calculable measures and then decide, based on their own experience, how far they want to expand the collaboration. A key factor for this undisturbed trust-building process: The platform completely avoids annoying advertising ads, so the editorial focus remains solely on the companies' expertise.

More information here:

• From visibility to trust: Your scalable path with Xpert.Digital