Google Is Now a Publisher: Germany’s War on the AI Answer Box

Germany's media regulator just classified Google's AI Overviews and Perplexity's chatbot as regulated publishers — not neutral search pipes — because synthesized answers sit above the link list, steer what journalism users see, and fall outside the EU Digital Services Act's liability shield.

On July 14, 2026, the Commission for Licensing and Supervision (ZAK) — the body representing Germany's 14 state media authorities — issued binding decisions finding both services subject to the State Media Treaty. The rulings are immediately enforceable. Google and Perplexity have one month to appeal.

What ZAK ruled:

• AI-generated summaries and chatbot answers are the provider's own content, not redistributed third-party material.
• The DSA exemption that shields platforms from illegal user posts does not apply to these outputs.
• Google's AI Overviews violate Section 109 of the State Media Treaty through discriminatory placement that buries traditional links — especially journalism — beneath Google's prose answer.
• Perplexity qualifies as a media intermediary when it selects and ranks source links alongside generated answers — but regulators flagged only a missing German representative and inadequate transparency disclosures so far, not the same discrimination finding Google faces.

Why placement matters:

ZAK chairman Dr. Thorsten Schmiege told reporters: "AI search engines and chatbots are content providers, and we will consistently apply German media law to them from now on."

On Google's product specifically, ZAK found that "the classic link overview is displayed as less findable compared to the AI answers, which are to be assessed as the provider's own content," per FAZ's reporting.

Schmiege added: "Anyone who controls whether content gets found through the selection and placement of links must make that transparent. Otherwise, diversity among journalistic and editorial outlets will disappear."

The Munich echo:

This is not Germany's first bite at the apple.

On May 28, Munich's regional court held Google directly liable for false AI Overview claims against two publishers — treating synthesized text as Google's speech, not cited websites'. ZAK's July order adds regulatory media-law obligations on top of that civil liability precedent.

Reuters noted the ZAK decision explicitly follows that court ruling and the broader European scrutiny of AI summaries.

Who brought the case:

• State media authorities in Hamburg-Schleswig Holstein and Berlin-Brandenburg initiated proceedings against Google and Perplexity to establish jurisdiction, per FAZ.
• ZAK's legal opinion underpins the enforcement orders — the first time German regulators have applied the State Media Treaty to AI search engines and chatbots at scale.

What Google said:

A Google spokesperson told Reuters the decision "fails to recognise how people's preferences when searching for information and the information ecosystem are changing."

The company said AI summaries "help users discover new content and ask follow-up questions" and confirmed it plans to appeal.

What Perplexity said:

Perplexity declined to comment on the ZAK decision. It told Reuters it complies with GDPR and holds SOC 2 Type II certification — neither of which addresses media-law classification.

Why the link defense fails — again:

Google includes source citations in AI Overviews. Regulators and courts treat that as insufficient.

A Pew Research Center study of 900 U.S. adults in March 2025 found users click a source link inside an AI Overview in just 1% of sessions. Overall click-through to any result drops from 15% to 8% when a summary appears.

Users read the answer inside Google's interface. The journalism underneath becomes optional.

The Preferred Sources fig leaf:

Google has rolled out a "Preferred Sources" feature letting users pick which outlets surface in AI answers — a pre-emptive compliance gesture regulators dismissed as inadequate.

Few users maintain custom source lists. Publishers who opt in get visibility; those who fight in court get buried.

Professors Jan Oster and Christoph Busch backed ZAK's view in a related legal opinion: AI search replaces ranked results with a single prose answer, cutting traffic to original reporting and putting journalism's funding at risk.

What changes for operators:

• Media-diversity duties now apply to AI answer engines operating in Germany — transparency on link selection, non-discrimination against editorial outlets, and formal representation requirements.
• Civil plus regulatory exposure stacks: Munich opened defamation liability; ZAK opens ongoing media-law enforcement.
• Enterprise AI products that synthesize answers under a corporate brand face the same reclassification logic — if you generate it and place it above third-party sources, you are not a pipe.

Convina's view: Munich told Google it owns the hallucination. ZAK just told it owns the business model. The DSA was built for platforms that host what users post — not for products that replace journalism with a branded paragraph and a 1% click rate. Germany is the first regulator to enforce that distinction at scale, and the appeal clock is ticking. Every enterprise shipping customer-facing AI summaries should assume the answer box is now a regulated publication — with diversity obligations, transparency duties, and no safe harbor for saying the links were still there.