Google AI Overviews Found Lacking in Accuracy and Potentially Harmful to Publishers
Researchers from Washington University in St. Louis have conducted an independent audit of Google's AI Overviews, revealing some concerning gaps in their accuracy and potential impact on publishers' revenue.
The study, set to be presented at the October 2026 ACM Internet Measurement Conference, analyzed 55,393 Google searches over 40 days. The results showed that AI Overviews appeared for 13.7% of all queries, but this rate varied sharply based on how users worded their search.
The researchers found that question-form queries and longer searches were more likely to trigger AI-generated answers. However, even when citing credible sources, the generated summaries did not always accurately represent them. The study revealed an 11% error rate, with 4% of claims directly contradicting the cited source.
The findings also suggested that Google's AI Overviews could threaten publishers' revenue. When users rely on these summaries for answers, they may lose the visit that generates advertising, subscription, or sales revenue. The researchers warn that this could have a significant impact on publishers who depend on advertising revenue.