Google Chrome's settings page made a quiet promise before its most recent update.
In Chrome version 147, under Settings > System > On-device AI, the description read: "To power features like scam detection, Chrome can use AI models that run directly on your device without sending your data to Google servers."
Privacy researcher Alexander Hanff confirmed the behavior using macOS kernel filesystem logs, and it has since been verified on Windows 11 and Ubuntu as well.
That argument already had a hole in it. Chrome 147's "AI Mode" pill in the address bar routes every query to Google's cloud, not the local Gemini Nano model.
Removing the "without sending your data to Google servers" language doesn't fix that contradiction. It just stops Chrome from making a claim it apparently isn't prepared to stand behind.
Google did not respond to Decrypt's request for comment on the settings change.
Chrome 148 is currently rolling out. Users still on version 147 will see the old text; anyone who has already updated will find the privacy claim gone.
