Chrome Quietly Drops Its Privacy Promise for On-Device AI – What You Need to Know
Google recently updated Chrome’s privacy policy, removing a commitment that on-device AI processing would stay entirely local. The change, first reported by Decrypt on May 7, 2026, and later picked up by Yahoo Tech, raises questions about how much data Chrome’s built-in AI features might now send to Google’s servers. If you use features like Smart Reply or automatic tab organization, here is what the update means and how to check your settings.
What Happened
Until this update, Chrome’s privacy policy explicitly stated that data processed by on-device AI models would never be sent to Google. That language has been removed. According to the reports, the new policy no longer guarantees that AI data stays on your machine. Instead, it allows for the possibility that some data may be transmitted to Google’s servers for processing or improvement of the AI features.
It is not yet clear exactly which features are affected or whether any data transmission has actually begun. The policy change itself is a shift in Google’s stated commitment, not necessarily an immediate change in how Chrome behaves. However, for privacy-conscious users, the removal of that promise is a red flag.
The original promise mattered because it distinguished Chrome’s on-device AI from cloud-based AI services. On-device processing was supposed to keep sensitive information—like what you’re typing or which tabs you have open—off the network. That assurance is now gone.
Why It Matters
Chrome’s on-device AI features include:
- Smart Reply (suggesting quick responses in text fields)
- Tab organizer (auto-grouping tabs by topic)
- Faster local AI models for certain tasks (e.g., translation, summarization)
If these features now send data to Google’s servers, then anything you type that triggers a Smart Reply suggestion, or the content of your tab groups, could potentially be transmitted. That is a significant change from the previous “local only” guarantee.
Privacy advocates have pointed out that this policy update is another example of how “on-device” can become a moving target. Companies often start with a strong local privacy claim and later dial it back as they find cloud processing more useful for model training or feature improvements.
For now, the change is in the policy language. But it opens the door for future data collection that users may not expect.
What Readers Can Do
You can disable Chrome’s on-device AI features to prevent any potential data sharing. Here is how to check your current settings in Chrome on desktop:
- Click the three-dot menu in the top-right corner and select Settings.
- Look for the AI section (in recent versions, it appears under “Experimental AI” or “Advanced” depending on your update level).
- Toggle off any features listed as “on-device AI” or “AI-powered features.” Options may include Smart Reply, Tab organization, Web AI, or Local AI model.
- If you see a setting labeled Allow data to be sent to Google for AI improvement, make sure it is turned off.
On mobile (Android and iOS), the exact path varies, but the AI settings are typically under Settings > AI. If you cannot find them, use the settings search bar and type “AI.”
Note that disabling these features may affect functionality—Smart Reply will no longer appear, and tab groups might not be automatically suggested. If you prefer the convenience, you can keep them enabled but be aware of the policy change.
Sources
- Decrypt, “Chrome Deletes Its Own Privacy Promise for Sneaky On-Device AI,” May 7, 2026.
- Yahoo Tech, “Chrome Deleted Its Own Privacy Promise for Sneaky On-Device AI,” May 7, 2026.
This is a developing situation. Keep an eye on your Chrome version updates and any further privacy notice changes from Google. For now, the safest course is to treat on-device AI as potentially off-device and adjust your settings accordingly.