When English ‘Best-in-Class’ becomes a German liability

Patrick Roye Published 9 June 2026 · Expanded canonical 14 June 2026 Read on LinkedIn

Rephrasing English superlatives for German style isn’t enough – the claim survives the rewrite, and so does the liability.

The legal framework has been there all along, yet most Anglo copywriters haven’t read it.

Under Germany’s §5 UWG – its law against misleading advertising – a superlative can be a factual claim. Thus terms like ‘best-in-class’ or ‘world-leading’ each carry a legal burden: the advertiser must be able to prove a substantial and durable lead.

In 2019, an OLG Frankfurt court ruled that advertising luggage with the English claim “World’s Lightest” on a German trade fair stand was misleading under §5 UWG – and asserted international jurisdiction over the foreign defendant even though they weren’t selling the product in Germany.¹ “World’s Lightest” was a measurable claim, and it failed the measurement: lighter cases were already on the market. The stand carried two other English slogans – “It takes anything you can throw at it” and “Trusted quality for generations” – and both survived. They made no measurable claim, whereas “World’s Lightest” did.

The legal exposure compounds a credibility problem that precedes any lawyer’s letter. The German B2B reader who enters scrutiny mode – triggered by unsubstantiated claims – becomes more skeptical of the claim and of the surrounding content.² The damage is hard to observe directly because readers rarely tell you why they disengaged. The credibility loss occurs before the content reaches legal review.

As important as this is, the standard pipelines aren’t built to catch it: the MT step simply renders the superlative as words, the QA pass weighs terminology and fluency, the non-specialist reviewer usually reads for accuracy and register – none of them asks whether the claim is provable. So the exposure ships through, unflagged.

Not every superlative fails, though. The English claim “Simply the best!” held up under §5 UWG. It survived because it was provable: the maker actually was the Stiftung Warentest winner (1.4 against the plaintiff’s 1.8), and a court that set that proof aside was overruled.³

“World’s Lightest” and “Simply the best!” faced the same law and reached opposite outcomes – the difference was whether they could be proven. A wet-razor claim may seem far from B2B software – but §5 makes no such distinction. A measurable superlative must be provable, whatever the audience.

Emerging risk · developing case law

The law is moving toward operator liability for AI output

The cases above are settled or higher-court authority. The development below is fresh and interim – weighted accordingly. It matters here because it supplies the mechanism the §5 argument implies but does not spell out: once an AI generates a claim rather than relays it, the operator is its author.

Landgericht München I, 28 May 2026 – Google liable for its AI Overviews

Two Munich publishers sued Google after its “Übersicht mit KI” (AI Overviews) falsely connected them to Betrugsmaschen, Abo-Fallen, and dubious business practices – merging information about other disreputable companies with the plaintiffs, connections present in none of the cited sources.⁴

The court classified Google as an unmittelbarer Störer – a direct interferer, the primary party responsible – rather than a mere mittelbarer Störer. Its reasoning: AI Overviews generate independent new substantive statements, summarizing and evaluating sources in the system’s own words rather than relaying third-party results. Only Google controls the AI and its algorithms, so Google answers for what the AI states. The court held that BGH precedent on classic search engines and autocomplete – which limits liability and imposes no proactive duty – was nicht übertragbar, not transferable, to AI Overviews.

Google raised three defenses, and the court rejected all three: that users can verify the sources themselves; that the overview is flagged as AI-generated; and that after-the-fact correction is possible. None released Google from liability. The remedy was an injunction only – Ordnungsgeld up to €250,000 per violation, with Google bearing roughly 80% of costs – and no damages. The ruling rests on an earlier line of reasoning from LG Frankfurt a.M. (September 2025, Az. 2-06 O 271/25), which had already raised AI-overview liability as a possibility.

Status, stated precisely. This is a preliminary injunction (einstweilige Verfügung), first instance, LG München I, 28 May 2026, Az. 26 O 869/26 – appealable and not final. As of mid-June 2026, Google had not decided whether to appeal. It is not settled German law, and it should not be read as such. It also concerns AI-generated search summaries making defamatory factual connections, not AI translation of marketing claims. The facts do not transfer; the principle does – once an AI authors a statement rather than relaying one, the operator is the author, and the operator’s liability attaches.

That is the bridge from the §5 provability test. An AI pipeline that renders an English superlative into an unprovable German claim is not relaying someone else’s words – it is producing the operator’s own statement, with the operator’s own liability attached. There is no intermediary to hide behind. A German court has now located that authorship in the operator.

The reach is continental

Substantiation is an EU-wide burden

The §5 exposure is not contained within Germany. In Lidl Belgium (C-356/04, 19 September 2006), the Court of Justice of the European Union confirmed that the substantiation burden for comparative and superlative advertising runs across member states.⁵ The enforcement pattern is consistent across jurisdictions: the Italian competition authority AGCM levied roughly €875,000 in penalties across 2018–2019 against TIM, Wind Tre, Fastweb, and Vodafone, assessed claim by claim; France’s ARPP framework allows penalties up to 10% of turnover; and Austria’s OGH addressed the same substantiation principle (4 Ob 223/22f, 2023). A claim that cannot be proven carries the same defect wherever in the EU it is published.

Sources

OLG Frankfurt am Main, 6 U 3/18 (2019), “World’s Lightest.” The court found the English-language superlative misleading under §5 UWG and asserted jurisdiction over the foreign exhibitor despite no German sales. Verified via openJur (2177330). ↩
The term scrutiny mode (Überprüfungsmodus) for the reader’s shift into heightened skepticism is attributed to Selders; it is not the author’s coinage. ↩
“Simply the best!” held up under §5 UWG because the maker held the Stiftung Warentest top score (1.4 against the plaintiff’s 1.8). A lower court that set that proof aside was overruled; the reversal was on procedural grounds, not a merits finding that the slogan was lawful in the abstract. BGH I ZR 17/09 (22 April 2010). ↩
LG München I, 26. Zivilkammer, 28 May 2026, Az. 26 O 869/26 (einstweilige Verfügung). First German court decision on liability for the AI Overviews feature; plaintiffs represented by Lausen Rechtsanwälte (per JUVE). Reported by Heise, Legal Tribune Online, and JUVE, June 2026. Preliminary, first-instance, appealable, not final. ↩
CJEU, Lidl Belgium GmbH & Co. KG v Etablissementen Franz Colruyt NV, C-356/04, 19 September 2006. Comparative-advertising substantiation obligations apply EU-wide. Italian AGCM penalties 2018–2019; France ARPP framework; Austria OGH 4 Ob 223/22f (2023) cited as the comparative enforcement pattern. ↩

When English ‘Best-in-Class’ becomes a German liability

Landgericht München I, 28 May 2026 – Google liable for its AI Overviews

Sources

About the author

This series