On July 16, German regulator ZAK classified Google’s AI Overviews and the Perplexity search engine as content providers, bound by media law. It is the first time a regulator has applied the interstate media treaty to AI search engines and chatbots. Both companies have one month to appeal.
Key Takeaways
- ZAK applies the Medienstaatsvertrag to AI Overviews and Perplexity, now treated as media outlets
- Obligations triggered: transparency, non-discrimination, protection of media diversity
- The ruling is immediately enforceable, with a one-month appeal window; Google is targeted for burying journalistic sources
Have an AI Sum Up This Article
ChatGPTA German Regulator Treats AI as a Media Outlet
ZAK, Germany’s commission for the licensing and supervision of media, issued its ruling on July 16. It classifies Google’s AI Overviews, the summaries generated at the top of the results page, and the Perplexity answer engine as content providers under the interstate media treaty, the Medienstaatsvertrag, in its Section 109. These are the same summaries Google is rolling out across Europe, with a France launch set for September 23.
Thorsten Schmiege, who chairs ZAK, laid out the logic plainly. AI search engines and chatbots are content providers, and Germany is now applying its media law to them consistently. The framing moves these tools from technical intermediary to publisher, responsible for what it puts out.
The step is a first. No German regulator had applied the Medienstaatsvertrag to AI search engines and chatbots before. The ruling turns a theoretical debate about the nature of these products into a dated legal obligation, with a named regulator and a precise statute behind it.
The reasoning aims at the core of responsibility. As long as a tool merely points to third-party sources, it stays a courier. The moment it writes its own synthetic answer, it becomes the author of a piece of content, and that author can no longer hide behind its sources.
Why the DSA Shield No Longer Protects
The European Digital Services Act, the DSA, protects platforms that distribute third-party content. ZAK finds that shield does not apply here. An AI-generated answer is independent content, not a simple redistribution of existing articles, so the protection falls away.
The result is a list of duties. Transparency about what is produced and how, non-discrimination in how content is surfaced, compliance with media diversity rules. These are a publisher’s classic obligations, applied for the first time to an automated summary.
Google is targeted on a specific charge. The regulator flags the placement of AI summaries at the top of the page, above the classic links, which pushes journalistic sources down. ZAK reads this as a breach of transparency and non-discrimination, the same logic that already led to a German ruling holding Google’s AI Overviews liable for false information.
Perplexity draws separate grievances. The regulator notes the lack of a designated representative in Germany and missing mandatory transparency disclosures. Two failings more administrative than the Google case, yet enough to fall under the same regime.
In concrete terms, content-provider status attaches a duty of traceability to every answer. A publisher must answer for what it puts out, correct an error, separate information from promotion. Applied to a generative summary, that duty forces the AI to make legible what it surfaces, on what criteria, and at whose expense. None of that was ever asked of a search index.
More articles on Horizon
- Google AI Mode Now Acts Inside Your Apps
- GPT-Red: OpenAI’s AI That Attacks Its Own Models
- Meta Lawsuit Claims AI Built the 8,000 Layoff List
What the Ruling Opens for Google and Perplexity
The ruling is immediately enforceable, and each company has one month to appeal. Google has already taken that road on nearby ground, when the company appealed the Munich ruling on its AI Overviews. The same legal argument will very likely return here.
If the appeal fails, the impact runs past this single case. An AI engine operating in Germany would have to document its display choices and stop disfavoring publishers in the page hierarchy. The constraint would land on the layout of the result itself, not only on its content.
For the press, the ruling offers fresh negotiating leverage. A publisher who feels buried under an AI summary now has a regulatory footing, not just a commercial argument. That footing did not exist a week ago. The balance of power with the platforms shifts by a notch.
Timing matters as much as substance. By setting the first case, ZAK hands other European regulators a template they can reuse, in a space where national media treaties stay flexible. An isolated precedent today can become a de facto norm tomorrow, and Brussels tends to watch this kind of member-state move closely.
For Google, the constraint lands at the worst moment. The company is pushing AI into every corner of search while a share of users already turns away, a move we tracked when DuckDuckGo gained 30% on the rejection of Google’s forced AI. Germany just added regulatory pressure to usage pressure, and the rest of Europe watches the precedent take shape.
Follow the story on Horizon.
