Chrome Drops Its Privacy Promise for On-Device AI – Here’s What to Do Now

A few weeks ago, Google quietly updated Chrome’s privacy policy. The change may seem small on paper, but for anyone who pays attention to how their data gets handled, it matters. The company removed a specific pledge that on-device AI features in Chrome would not send any data to Google’s servers. If you’ve been relying on that promise to use features like smart text selection or page summarisation without worrying about your information leaving your machine, it’s time to take a closer look.

What Actually Happened

Until early May 2026, Chrome’s privacy policy stated that on-device AI models would run locally on your device and “no data is sent to Google” during those operations. That language is now gone. The update was first spotted and reported by Decrypt, and later confirmed by Yahoo Tech.

Google has not explained exactly why the change was made. In response to media inquiries, the company said little more than that the policy now reflects how the features actually work. But that explanation raises a question: if on-device AI was genuinely running without sending data to Google before, what changed? Either the earlier promise was inaccurate, or the data collection is new.

Why It Matters for Your Privacy

On-device AI is supposed to be a privacy-friendly approach. By processing tasks like translation, image description, or topic classification directly on your computer, it avoids shipping your browsing history, document contents, or personal notes to remote servers. That’s why the original promise was reassuring to privacy-conscious users.

Removing that promise doesn’t automatically mean Chrome is now uploading everything to Google. But it does mean the company no longer guarantees that your data stays local. Without that guarantee, there’s a risk that the AI features could send information – perhaps metadata, perhaps more – back to Google’s servers for model improvement or other purposes. And because the change was made quietly, most users never got a chance to consent or object.

This is especially relevant because Chrome’s AI features are being rolled out more broadly. Features like “Help me write”, automatic page summarisation, and contextual suggestions are becoming harder to avoid. If you use Chrome regularly, your browser might already be running these tools without explicit notice.

How to Protect Yourself

You don’t have to wait for Google to explain further. Here’s what you can do now.

1. Disable on-device AI features in Chrome

Google has put some controls under chrome://settings/ai. Type that address into your Chrome address bar and hit Enter. You’ll see a list of AI-powered features. For each one that you don’t actively need, toggle it off. Pay special attention to anything labelled “on-device AI” or “local AI” – those are the ones that no longer have the privacy promise.

Common features to check:

  • “Help me write”
  • “Page summarization”
  • “Smart selection”
  • “Live captions” (though this one often runs locally anyway)

Turning them off won’t break your browser. It may reduce convenience, but it removes the uncertainty about data flow.

2. Review your Google account privacy controls

Even if you disable the AI features, Chrome still sends other data to Google (search queries, sync data, crash reports). Visit myaccount.google.com and go to “Data & privacy”. Turn off “Web & App Activity” if you don’t want your browsing linked to your account. Also disable “Ad personalization” to limit profiling.

3. Consider an alternative browser

If you want stronger guarantees, no browser is perfect, but some are more transparent than others. Firefox does not bundle on-device AI features that phone home. Brave includes some AI tools but gives you clear toggles and does not send data to third parties by default. Even Microsoft Edge, after its recent privacy policy updates, offers more granular controls for its AI features than Chrome currently does.

Switching isn’t for everyone, but it’s worth testing for a week to see if you can live without Chrome-specific ecosystem features.

Looking Ahead

Google hasn’t said whether the data collection is actually happening or just not ruled out. That ambiguity is the problem. Companies that care about user privacy do not silently remove protective language from their policies. If you value knowing where your data goes, this is a clear signal.

The good news is that you can act today. The settings are there – you just have to find them. And if Chrome keeps moving in this direction, the best long-term option may be to use a browser that treats local AI as a real privacy feature, not a marketing claim.

Sources

  • “Chrome Deletes Its Own Privacy Promise for Sneaky On-Device AI” – Decrypt, May 7, 2026.
  • “Chrome Deleted Its Own Privacy Promise for Sneaky On-Device AI” – Yahoo Tech, May 7, 2026.
  • Chrome privacy policy (archived versions) via Internet Archive for comparison.