# Google Help Page Spells Out How Signed-In History Trains AI Overviews

Published: 2026-07-24T19:58:07.286Z · Source: Google AI Overviews documentation (https://support.google.com/websearch/answer/14901683)
Source date: 2026-07-24
Flags: documentation change bounded to Jul 23–24 UTC — AEO Wire watch
Entities: ai-overviews, google

Google's AI Overviews support documentation states that signed-in users' search interactions, alongside feedback and human review, are used to improve Search's generative AI models; users can stop future collection via My Activity, but Google says it may still use aggregated, anonymized search data for model improvement even after opting out. AEO Wire spotted the language through a routine documentation watch: it was absent from our snapshot of the page taken at 23:13 UTC on July 23 and present in our next snapshot at 01:15 UTC on July 24, and we confirmed it against the live page. Google does not date individual help-page edits, so the exact edit time is unconfirmed; our snapshots bound the language's appearance to that roughly two-hour window.

Why it matters: The training-data disclosure moved from policy abstraction to plain help-page language — quietly, in a roughly two-hour window our watchers caught.

## What this answers

**Does Google use my search history to train AI Overviews?**

Google's support page says signed-in users' search interactions — with feedback and human review — improve Search's generative AI models; you can stop future collection via My Activity, though Google says aggregated, anonymized data may still be used.

**How do I opt out of Google using my searches for AI training?**

Per the page, turning off collection in My Activity stops future Search history from being saved, though previously saved history is unaffected unless deleted.


Canonical: https://anythingengineoptimization.com/item/2026-07-24-google-help-page-spells-out-how-signed-in-history-trains-ai-overviews/
From Anything Engine Optimization (AEO Wire) — https://anythingengineoptimization.com/ · Standards: https://anythingengineoptimization.com/standards/