German Court Holds Google Liable for AI Overview Errors

The Munich I Regional Court has ruled that Google is liable for content generated by its AI Overview feature in the same way as for its own statements, rather than as a neutral intermediary passing along search results. It is one of the first rulings of its kind in Europe and could reshape how the company bears legal responsibility for its AI's mistakes.

What the court found

The case involved two Munich-based publishers whose names the court did not disclose. They complained that Google's AI Overview feature attributed to their companies information about questionable business practices that actually belonged to other, genuinely dishonest entities. The AI system combined data from different sources and companies, creating connections that did not exist in any of the materials cited as the basis for its answer.

Central to the ruling was the distinction between a conventional search result and AI-generated content. The court found that AI Overview does not simply index and paraphrase material already available online, but produces a new, autonomous statement by selecting, synthesizing and evaluating information from multiple sources at once.

Displaying AI-generated results constitutes its own, separate content that can be attributed to Google - Munich I Regional Court

How this departs from existing rules

Until now, search engines have relied on limited liability for third-party content they merely link to. The Munich court found that this principle does not automatically extend to a generative format. Because AI Overview independently formulates sentences and conclusions, ordinary personal-rights and defamation standards apply to it, just as they would to any other content publisher.

The court ordered Google to remove the disputed content and to prevent the publication of similar false claims against the plaintiffs. Failing to comply with the ban carries a fine of up to 250,000 euros. Google is also covering about 80 percent of the proceedings' costs, which courts typically treat as a sign of which side had the stronger case.

Google's response and what comes next

Commenting on the case, Google Poland disagreed with the ruling and announced it would appeal, stressing that the decision is not yet final.

The case concerns only isolated shortcomings, not the underlying mechanism - Google Poland

The company argues that the vast majority of AI Overview's generated answers are factually correct, and that this incident involves a single, isolated malfunction of the feature. The ruling was issued through an interim injunction, an expedited procedure intended to prevent further harm before the court rules on the merits of the case. A final decision may not come until the appeal has been resolved.

Implications for the market and Poland

The ruling arrives as AI Overview and similar generative summary features face growing criticism over factual errors. Analysts estimate the system gets things wrong in roughly one in ten cases, and deploying newer models improves accuracy by only a few percentage points. For publishers and companies harmed by false summaries, the Munich ruling points to a real legal avenue: a personal-rights lawsuit rather than futile complaints filed with Google itself.

For Polish companies and law firms, the ruling carries practical significance even though it was handed down in Germany and does not bind Polish courts. It nonetheless signals a possible direction for European case law on platform liability for generative content, at a time when Poland is working on regulations that would clarify the legal status of AI systems as products. Companies monitoring their presence in AI-generated search results now have a precedent they can cite in disputes with major platforms.