Chrome Quietly Removed Its Privacy Promise for On-Device AI: What It Means for You

In early May 2026, several news outlets noticed something that had quietly changed in Google Chrome’s documentation. The privacy promise that on-device AI features do not send data to Google servers had been removed. If you use Chrome’s built-in AI tools—things like smart tab grouping, writing assistance, or image generation—you may now be sharing your browsing habits with Google more than you expected.

What Happened

For years, Google maintained a line in its Chrome privacy policy stating that on-device AI processing “does not send any data to Google servers.” That language is gone. As reported by Decrypt and Yahoo Tech around May 7, 2026, the deletion wasn’t announced; it simply appeared in updated policy text.

Google has not officially commented on why the change was made, but the implication is clear: some AI features that previously promised to keep everything local may now upload data to cloud servers for processing. The exact details of what data is sent and under what conditions remain vague—Google’s current documentation no longer draws a firm line between on-device and server-based AI.

Why It Matters for Your Privacy

This change matters because “on-device” was a key privacy selling point. Many users chose Chrome specifically because Google said AI processing wouldn’t touch their servers. Now that guarantee is gone.

  • Exposure of browsing patterns: Even if only anonymous snippets are sent, aggregated over time they can reveal what you read, which sites you visit, and how you interact with content.
  • Loss of clear boundaries: Without a clear promise, you have to trust that Google won’t use this data for profiling or training. Google’s business model relies on data, so that trust may be misplaced.
  • Precedent for future features: If a privacy promise can be removed silently for one feature, it could happen for others (e.g., Chrome’s suggested searches, spelling correction, or location services).

It’s also worth noting that this coincides with Chrome’s push to integrate more AI features, which increases the surface area for potential data collection.

What You Can Do Right Now

You can’t control what Google documents say, but you can control which features are enabled in your browser.

  1. Open Chrome settings. Click the three-dot menu (top-right corner), then choose Settings.
  2. Go to Privacy and Security. In the left sidebar, click Privacy and Security.
  3. Find the AI section. Look for a heading like AI or On-device AI (exact wording varies by version). It may also appear under Advanced.
  4. Disable AI features. You’ll see toggles for things like “Smart tab grouping,” “Help me write,” “Autofill with AI,” and “Image generation.” Turn off any you don’t need or simply toggle off “Enable on-device AI” if that top-level switch exists.
  5. Review permissions for site data. Go to Privacy and Security > Site Settings > Additional permissions. Any AI tools that require website content should be set to “Block” or “Ask.”

If you prefer a more private browser altogether, consider switching to Firefox (which has strong tracking protection and is working on its own privacy-first AI features) or Brave (which blocks most data collection by default). Both still let you use AI tools with more transparency about when data leaves your device.

What to Watch For

This incident shows how quickly privacy promises can be revised without public discussion. Keep an eye on:

  • Updates to Chrome’s privacy policy: bookmarks.google.com/chrome/privacy/ (or the version for your region). Compare old and new language using site archives like Wayback Machine.
  • Any official response from Google about what changed and why.
  • Reports from security researchers who test whether Chrome’s AI features actually call home.

For now, the safest approach is to assume that any AI-powered feature in Chrome could send data to servers unless you explicitly disable it. This might be inconvenient, but it’s better than assuming a promise that no longer exists.

Sources: Decrypt, “Chrome Deleted Its Own Privacy Promise for Sneaky On-Device AI” (May 7, 2026); Yahoo Tech coverage of the same report (May 7, 2026). These outlets first identified the removal of the privacy assurance from Chrome’s documentation.