A Court Just Ruled the AI's Words Are Yours.

The Regional Court of Munich issued a temporary injunction barring Google from repeating false claims its AI Overviews generated about two Munich publishers — wrongly tying their names to scams, subscription traps, and “dubious business practices.” The catch: those claims appeared nowhere in the linked sources. Google's own AI invented them.

The court's reasoning is the earthquake. Traditional search engines are intermediaries — they point you elsewhere, and safe-harbor law shields them. But an AI Overview doesn't point. It rewrites, structures, and synthesizes information into a brand-new, self-contained statement. So the court ruled those statements are Google's own words — and Google is directly liable for them.

Then it closed the last exit. Google's fallback — that users should verify AI answers themselves — was rejected outright. The judges noted that if every answer requires manual verification, the AI summary has no value at all. You can't sell a machine as an authority and then disclaim it as a guess. Google was ordered to stop and stuck with 80% of the costs. It is appealing.

The principle is now on the board: when your AI speaks, you spoke. And it doesn't stop at search — it lands on enterprise copilots, research assistants, every tool that turns sources into answers. HITL Score: 14/100 — human oversight at deployment 4/25 (AI Overviews ship answers to billions with no pre-publication human review; the system synthesized defamatory claims that existed nowhere in the source material), ongoing monitoring 3/25 (errors surfaced only because the victims found them and lawyered up — the monitoring was the lawsuit), incident response 4/25 (the response wasn't remediation, it was an appeal; the court had to order Google to stop repeating the claims), accountability 3/25 (the entire defense rested on “we're just an intermediary / users should verify,” and the court demolished both).