Germany Rules Google's AI Overviews and Perplexity Subject to Media Law in Landmark Decision

Germany's media regulator ZAK has for the first time classified AI search engines and chatbots as content providers, ruling against Google and Perplexity. ZAK chairman Thorsten Schmiege stated that AI search and chatbots are content providers, and German media law will apply consistently to them.

ZAK argues that the Digital Services Act's exemption for third-party content does not apply to AI-generated content because it is the provider's own content, not redistributed third-party material. This logic aligns with a recent Munich court ruling that treated AI-generated text as independent content and held Google liable for false information.

ZAK accuses Google of violating transparency and anti-discrimination rules. Google's AI Overviews are placed at the top of search results, crowding out traditional links, especially news sources. The regulator believes that since AI Overviews are Google's own content rather than neutral search results, this crowding out constitutes prohibited discrimination.

However, link visibility may not be the only issue. Research shows that users rarely click on source links once they feel their question is answered. Google argues these studies are flawed but has not published contrary data.

In Perplexity's case, the regulator currently only notes its lack of a German representative and transparency disclosures. Although theoretically the same concerns should apply because the service operates similarly, Google's influence is greater.

The regulator also considers AI services to play an intermediary role. When a chatbot includes third-party content as sources or link lists, it shapes users' ability to find that content, which meets the criteria for a media intermediary and triggers transparency rules aimed at protecting media diversity.

Relevant legal opinions support the regulator's view. Professors Jan Oster and Christoph Busch argue that adding AI to search engines changes how people access information, from a list of links to a single prose answer, reducing traffic to original sources and endangering journalism funding. They suggest creating a separate category for AI search engines under state media law and establishing rules to protect media diversity.

Google has prepared in various ways, including launching the 'Preferred Sources' feature. But this is seen as a 'fig leaf' allowing Google to argue in court that users can choose sources. However, few users will maintain custom source lists, and publishers are queuing up to promote the feature. For Google, it's a free pass to replace original sources in AI replies with providers who cannot or will not fight back in court.