A Munich court called AI Overviews Google's own content. A Berlin court called them search results. The fight over which framing wins will shape who pays when the AI summary box at the top of Google gets it wrong.
Google is appealing a Munich Regional Court decision that, for the first time, treats its AI-written answers at the top of Search as Google's own content rather than as neutral search results, making the company directly liable when those answers are wrong. The appeal sets up a rare, same-month, opposite-direction ruling from a Berlin court that reached the other conclusion on the same product, and it puts at stake a question that will ripple beyond Germany: when a search engine publishes a generated summary, is the publisher of the answer Google, or is Google merely a host for an algorithmic pull-quote from the open web?
The Munich case, decided in late May 2026, grew out of a concrete injury. According to The Decoder's report on the appeal, Google's AI summary box at the top of search results falsely linked two Munich-based publishers to fraud schemes. The court treated those summaries as standalone content, the kind of statement a publisher owns and vouches for, not as a neutral list of links. That classification made Google the responsible party for the error rather than a passive conduit for third-party pages.
A few weeks later, a Berlin court looked at the same AI product and reached the opposite conclusion. There, the court classified the AI summary box as another form of search result, which keeps Google in a limited-liability posture as an indirect contributor to what third-party pages say. Google is expected to lean on the Berlin ruling in its appeal, according to The Decoder, because the second ruling undercuts the doctrinal premise that Munich used to make Google the author of the answer.
A Google spokesperson told reporters the case is about specific, narrow errors in the AI summary, not about how the feature itself displays web content. The company has not, however, drawn the line between the two. Without a public definition of which factual mistakes cross from "search result" to "authored content," the appeal asks courts to keep limited liability for a product Google presents to users as its own AI-generated answer rather than a curated list of other people's pages.
The split matters because the AI summary box is the new front page of Google Search for many queries, and the question of who stands behind it determines who absorbs the cost of a wrong answer. Under the Munich framing, the burden lands on Google. Under the Berlin framing, it shifts toward the publishers whose work the summary is drawing from, and toward the users who read the summary as a Google-stamped answer. Publishers whose names or work are misrepresented have one set of legal tools in Munich and a thinner one in Berlin, and the answer will not settle until a higher German court, or a court elsewhere in the EU, picks a frame.
Both rulings are first-instance and apply only in Germany. The Munich decision is the one under appeal, and the Berlin decision is a competing data point Google is putting on its side of the scale. The next move is Google's appellate brief, and the watch item is whether the higher court treats the two rulings as a doctrinal fork worth resolving or as parallel trial-court outcomes that can coexist.