AI Overviews: when the summary is false, Google is at fault, according to German justice

A historic verdict issued by the Munich Regional Court has ruled that Google is directly responsible for false statements contained in its "AI Overviews", the AI-generated summaries displayed in search results. The court classified such summaries as "Google's own content", rejecting the company's argument that the responsibility for fact-checking would fall on users.

The ruling stemmed from a lawsuit (number 26 O 869/26) in which Google's AI Overviews had erroneously associated two publishers based in Munich with scams, illegal business practices, and subscription "traps". The court found that the AI had mixed information about genuinely questionable companies with that of the plaintiffs, creating nonexistent links in the original sources. The judges clarified that AI Overviews works differently from traditional search results: as we know, the latter index and redirect to third-party content, while Google's AI rewrites and reprocesses the results "with its own words and structure", formulating "independent, new, and substantial statements".

Google's Defense in the AI Overviews Case in Germany

In this specific case, the AI started with definitive claims, then built its own narrative with summaries and alerts about alleged scams. As Google developed and provided this technology to users, the court concluded that it is the only entity that has influence over its functioning and algorithms, thus making it fully responsible for what is produced.

Google attempted to defend itself by arguing that users could have verified the linked sources to ascertain the accuracy of the AI summary, and that they were generally aware of the need not to blindly trust information generated by artificial intelligence. However, the court rejected this line of defense, stating that the ability to refute a statement through further research does not exempt one from responsibility.

The ruling drew a parallel with press law, where publishers are held accountable for independently understandable "teasers", even if readers never consult the full article. Additionally, the court highlighted a protection gap: if Google were only responsible for blatant violations, victims would not have an effective legal remedy, given that third parties had not made the incriminating statements at all. The court also stated that the opinion of an artificial intelligence is not "the expression of a belief acquired by the people who express it, but the result of an algorithm". Offering AI-based search is "primarily an expression of Google's business activities" and only "at most a secondary expression of an interest in being able to freely express one’s opinion and beliefs".

Therefore, the privacy interests of the plaintiffs took precedence over those of Google, especially considering that the contested claims were based on untruthful facts.

Google lost on nearly all fronts, with only two minor requests rejected, and has been ordered to cover 80% of the legal costs. The ruling could set a significant precedent for the liability of all AI service providers that paraphrase content from the web, including names like ChatGPT, Claude, or Perplexity. As of the time of writing, Google has yet to comment on the news.