Google Removed Chrome’s Privacy Promise for On-Device AI: What to Do Now

If you’ve been relying on Chrome’s on-device AI features because they were marketed as privacy-safe, you might want to double-check the fine print. Earlier this week, news broke that Google quietly deleted a key commitment in its documentation — one that explicitly stated on-device AI would not compromise user privacy. The change was first reported by Decrypt on May 7, 2026, and later confirmed by Yahoo Tech. For privacy-conscious users, this raises a straightforward question: what exactly changed, and what should you do about it?

What Happened

Chrome’s on-device AI features, such as the built-in “Help me write” tool and other local processing capabilities, were introduced with a promise that everything would happen on your device. Google originally wrote that these features “will not compromise your privacy.” That language has now been removed from relevant help pages without any public announcement. The Decrypt article noted that the removal appears to be recent, and that the new phrasing is vaguer — it no longer gives the same clear guarantee that data stays local.

It’s worth noting that we don’t have a full timeline of when the wording changed, nor has Google published an explanation. But the difference is real. Where Chrome once said on-device AI wouldn’t affect your privacy, it now says something less specific, leaving room for potential data collection or model updates that might require a connection.

Why It Matters

On-device AI has been one of the few areas where tech companies could honestly say “your data never leaves your device.” That promise was a major reason many users felt comfortable enabling these features. Removing that commitment — especially without a clear replacement — erodes trust. Even if the actual technical behavior hasn’t changed yet, the door is now open for future updates that could send usage data, feature requests, or telemetry back to Google’s servers.

The lack of transparency is the real issue. Chrome users didn’t sign up for a vague policy shift. If you enabled on-device AI based on the original privacy pledge, you now have a right to know what data, if any, is being processed remotely.

What You Can Do

Here are concrete steps to take right now, from least to most involved.

1. Check your current AI settings.
Open Chrome and type chrome://settings/privacy into the address bar. Scroll to the “AI” or “On-device AI” section. Look for toggles labeled something like “Help me write” or “AI features that run locally.” Depending on your version, you may see a note about how data is handled. If the description is vague or doesn’t explicitly say “no data leaves your device,” consider turning those features off.

2. Disable automatic download of AI models.
Some Chrome AI features need periodic model updates. In the same settings area, check if there’s an option to prevent automatic downloads. If you’re on a metered connection or simply don’t want extra network activity, turn that off.

3. Use a browser with clearer privacy policies.
If you’re uncomfortable with Chrome’s shift, you can switch to an alternative that either doesn’t have AI features or is more transparent about them. Firefox, Brave, and Vivaldi all offer good privacy protections. Firefox, for instance, has been adding local AI features with explicit on-device promises that they’ve said they’ll maintain.

4. Install a privacy extension.
For those who want to stay on Chrome, extensions like Privacy Badger or uBlock Origin can help block some data collection, though they won’t necessarily interfere with Chrome’s own AI telemetry. You can also use the extension “Chrome Cleaner” to audit what kind of network requests Chrome makes.

5. Keep an eye on updates.
Google hasn’t commented on the change yet. Follow credible privacy news sources (like the ones that broke this story) to see if they issue a clarification. In the meantime, treat on-device AI in Chrome as you would any cloud service — assume data could be shared unless proven otherwise.

Sources

  • Decrypt, “Chrome Deleted Its Own Privacy Promise for Sneaky On-Device AI,” May 7, 2026.
  • Yahoo Tech, confirmation of the same report, May 7, 2026.
  • Chrome settings documentation (current version as of May 2026).

This situation is still developing. The immediate takeaway is simple: don’t trust promises that have already been taken down. Review your settings now, and make your own call.