Analysis Finds That Google’s AI Overviews Are Providing Misinformation at a Scale Possibly Unprecedented in the History of Human Civilization

A new Oumi analysis commissioned by The New York Times found Google’s AI Overviews were about 91% accurate with Gemini 3 and 85% with Gemini 2.

Because Google handles trillions of searches a year, even a small error rate could still create millions of wrong answers at scale.

The report warns that AI search summaries can spread misinformation widely, especially when users trust them without checking sources.

Google disputed the study, saying it does not reflect typical searches and that higher model accuracy does not guarantee better grounding.