German official says Google and Perplexity must follow media laws

Germany’s media watchdog said Google’s AI Overviews and Perplexity AI are content providers under the country’s media law on Tuesday. This means that the two most popular AI search engines now have to follow the same rules as regular publishers and are directly liable for the content generated by their AI systems.

The decision was made by ZAK, the Commission for Licensing and Supervision, which is made up of 14 German state media. ZAK found that the companies running AI created news summaries and chatbot responses are the authors. That’s a critical distinction because it implies AI products can no longer hide behind the legal shield they have relied on in the past.

“AI search engines and chatbots are content providers,” ZAK Chairman Thorsten Schmiege said in a statement. “From now on, we will always apply German media law to them.”

In those cases, the Digital Services Act’s responsibility exemption is not applicable, the regulator said. That immunity typically shields platforms from liability for user created illicit content.

German court holds Google responsible for AI Overview content

The decision by ZAK came after a court in Munich said that Google was directly responsible for the allegedly false statements that its AI Overview feature made. The court saw the summaries as Google’s own work, not as repackaging information from other sources.

That reasoning is the same as what ZAK applied to the broader issue of media law. Once an AI system is the author of what appears on screen, the company behind it bears the obligations and risks associated with publication.

ZAK says that Google’s AI Overviews show up at the top of search results, pushing down traditional lists of links. The regulator thinks this is unfairly hurting third-party media outlets that depend on that traffic.

ZAK says that tools like Perplexity change the news that people read by letting users pick and show sources, links, or suggestions along with their own answers. On that basis, the regulator stated that such services could qualify as media intermediaries, a category with rules designed to protect media plurality.

The German move comes as global pressure grows for formal oversight of frontier AI systems. According to Cryptopolitan, Google DeepMind CEO Demis Hassabis recently called for a US watchdog, similar to Wall Street regulator FINRA, to screen the most powerful AI models before they are released. Anthropic CEO Dario Amodei has advocated for a more powerful agency, similar to the FAA, with the authority to prohibit unsafe models.

These proposals address safety concerns with the underlying models. Germany’s regulator is looking into how AI products distribute news and who is responsible if the output is incorrect.

If you're reading this, you’re already ahead. Stay there with our newsletter.