German media regulators have issued what appears to be the first formal media-law rulings against major AI search products, declaring that Google AI Overviews and Perplexity should be treated as content providers rather than neutral distributors of third-party material. The move matters well beyond Germany because it challenges a core assumption behind AI search: that generated answers can be regulated like search results or platform hosting.

According to reporting by The Decoder, Germany’s Commission for Licensing and Supervision, known as ZAK, found that the services fall under the country’s State Media Treaty. The regulators’ position is that AI-generated answers are the companies’ own content, which means the usual liability shield for platforms under the Digital Services Act does not apply in the same way. Both Google and Perplexity reportedly have one month to appeal.

For AI builders, enterprise buyers, and publishers, the significance is immediate. If regulators in Europe begin treating AI answer engines as editorial products instead of neutral intermediaries, vendors may face tighter rules on transparency, ranking, source treatment, and legal liability. That would affect not only consumer search, but also any enterprise AI system that synthesizes outside information into a single response.

What Germany’s ruling actually says

The Decoder reports that ZAK has formally ruled against services from Google and Perplexity under Section 109 of the State Media Treaty. In Google’s case, regulators argue that Google AI Overviews receive privileged placement above standard search links and therefore push traditional results, including journalistic sources, into the background.

That distinction is central to the ruling. Regulators are not just objecting to AI-generated summaries in the abstract. They are reportedly arguing that once Google presents a synthesized answer in the most prominent slot, it is no longer merely organizing external content the way a classic search engine does. Instead, it is publishing its own response while still controlling which outside sources remain visible beneath it.

According to The Decoder, ZAK also says AI systems can qualify as media intermediaries when they include third-party sources or link lists, because they influence which content users discover. In the German media-law framework, that triggers transparency obligations intended to protect media plurality and prevent discriminatory treatment of publishers.

Perplexity appears to be in a different procedural position. The Decoder says regulators have so far flagged the company for lacking a designated representative in Germany and for missing transparency disclosures. The article suggests broader concerns could apply to Perplexity as well because its product similarly returns synthesized answers and selected sources, but the reported ruling against Perplexity currently appears narrower than the one aimed at Google.

Why the liability shield question matters

The legal hinge in this case is whether AI-generated output is treated like hosted third-party content or like newly created content from the service itself. The Decoder reports that German regulators believe the Digital Services Act liability exemption does not cover these AI responses because the outputs are independently generated, not merely republished.

That reasoning aligns, according to the same report, with a recent Munich court decision that treated AI-generated text as independent content containing new and substantive statements assembled from multiple sources. The Decoder says that court held Google liable for false claims and that Google plans to appeal. If that line of reasoning spreads, AI product companies may need to defend generated answers as direct product output rather than fallback to platform-style immunity.

This is a meaningful shift for the broader AI market. Many generative products are designed around synthesis: retrieve documents, combine them, and answer in natural language. If regulators say the synthesis itself creates new publisher liability, then the compliance burden grows for search products, research assistants, customer support copilots, and internal enterprise AI tools that summarize external data.

For enterprise teams, that does not automatically mean such tools become unusable. It does mean procurement, risk, and legal teams may start asking different questions: who is responsible for generated claims, what disclosures appear to users, how are sources ranked, and can the vendor explain why one publisher or document was surfaced over another.

The pressure point is not only accuracy, but traffic and visibility

Germany’s move is also a direct challenge to the economics of AI search. The Decoder frames regulators’ concern as partly about visibility for journalism. If users receive a complete answer from Google AI Overviews, the incentive to click through to the original source falls, even if links are still displayed.

That matters because traditional web search has long sent traffic outward, while answer engines increasingly absorb the user’s attention on-platform. The Decoder notes studies suggesting users rarely click source links once they feel their question has already been answered, although it also reports that Google disputes those studies as flawed and has not, at least in the cited reporting, released alternative data to rebut the traffic concern.

This is why the German argument is bigger than one feature placement dispute. Regulators appear to be saying that AI search changes the structure of information discovery itself. Instead of users comparing multiple links, they are presented with a single prose answer selected and framed by the model and the product interface. From a media-law perspective, that shifts power over visibility and attribution toward the AI service.

Google has already introduced Preferred Sources, a feature mentioned by The Decoder, that allows users to influence which sources appear. But the article treats that as an incomplete answer to the underlying complaint. The regulatory issue is not simply whether a user can personalize sources; it is whether the platform’s default answer format systematically displaces original reporting and whether the service is transparent about that ranking logic.

Evidence, claims, and what remains uncertain

The strongest factual claims in this story come from The Decoder’s report on the ZAK rulings. Tech Times also characterized the development as a first-of-its-kind media ruling and said Germany had stripped AI search of an EU liability shield, but the full text available in the source material is limited.

Several important points should therefore be treated carefully. First, while The Decoder says the rulings are immediately enforceable and that Google and Perplexity have one month to appeal, the source material here does not include the full regulator decisions or public filings from the companies. Second, the exact legal scope of the Perplexity ruling is less detailed in the evidence than the Google case. Third, the broader claim that this is the world’s first such ruling is reported by the media sources in this cluster, but readers should still watch for the publication of primary legal documents or comparable actions elsewhere.

The article also includes interpretation from a legal opinion by Professors Jan Oster and Christoph Busch, as cited by The Decoder. Their view, according to that report, is that AI search engines deserve a separate category under state media law because they reshape how people find information and could threaten media diversity and publisher economics. That is expert commentary, not itself a binding legal determination.

Google’s side is only partially reflected in the available evidence. The Decoder says Google disputes some click-through studies and is preparing legal responses, including an appeal in the related Munich case. No fuller comment from Google is included in the materials provided here. Likewise, the evidence set does not include a direct response from Perplexity.

What this means for AI builders and enterprise buyers

For product teams building AI agents, retrieval systems, or AI search interfaces, Germany’s approach creates a practical design warning. The more a product synthesizes information into an authoritative answer, the harder it may be to argue that it is a neutral conduit. That raises the value of source transparency, answer traceability, jurisdiction-specific compliance, and controls over ranking logic.

This will be especially relevant in enterprise AI deployments that mix public web information, licensed content, and proprietary data. Buyers may ask vendors whether their systems can explain source selection, preserve meaningful citations, and separate generated interpretation from underlying documents. They may also ask whether regional compliance settings differ across the EU.

For consumer-facing search, the competitive implications are sharper. Google Search has the reach, distribution, and default placement to make any adverse ruling consequential. Perplexity, while smaller, is an important signal case because its core product is AI-native answer search rather than a legacy search engine adding generative layers. If the same legal logic applies to both, the regulatory category may attach to product behavior, not company size.

The story also lands as European scrutiny of AI systems continues to widen from model training and safety into interface design, discoverability, and market power. In that sense, Google AI Overviews and Perplexity are becoming test cases for how regulators handle AI-generated mediation of information rather than only the models underneath.

What to watch next

The first signal is whether Google or Perplexity appeals and on what grounds. An appeal would help clarify whether the companies intend to fight the classification of generated answers as their own content, the media-intermediary logic, or the specific transparency findings.

The second is whether German regulators publish fuller reasoning or remedies. For example, it will matter whether compliance can be achieved through disclosures and process changes, or whether regulators ultimately push for interface changes that reduce the prominence of AI answers.

Third, watch whether other EU member states or national media regulators adopt similar reasoning under their own laws. If they do, this could become a broader operating constraint for AI search across Europe rather than a Germany-specific anomaly.

Finally, publishers and enterprise buyers will be watching for data. If vendors such as Google provide stronger evidence on click behavior, source attribution, or publisher traffic outcomes, that could shape both legal arguments and product design.

Creati.ai perspective

Germany’s move is notable because it targets the product layer of generative AI, not just the training stack. Regulators are focusing on how AI answers are presented, how they reorder attention, and who bears responsibility for the resulting claims. That is a more concrete line of attack than abstract debates about AGI risk or model openness, and it is one that directly affects shipping products.

For builders, the lesson is straightforward: if your system produces a polished answer that users treat as final, expect regulators to view that answer as yours. In AI search and enterprise AI alike, citation quality, disclosure, and explainable ranking are no longer just UX features. They are becoming part of the legal architecture of the product.