Chrome Just Quietly Deleted Its Privacy Promise for On-Device AI — Here’s What to Do

Google has removed a long-standing privacy promise from Chrome that assured users on-device AI features would not send data to its servers. The change, first reported by Decrypt on May 7, 2026, happened without a public announcement or clear notification to users. If you use Chrome’s built-in AI tools—such as smart reply, tab organization, or image description generation—this shift could affect how your data is handled.

What Happened

For years, Chrome’s documentation stated that on-device AI models would process everything locally, meaning no data would leave your device. That language has been deleted. According to Decrypt’s report, the company now uses more ambiguous wording that no longer guarantees local-only processing. The exact timeline of the removal is unclear, but the promise that once formed a key part of Chrome’s privacy messaging is simply gone.

Google has not issued a statement explaining the rationale. However, the move aligns with broader industry trends: companies like Microsoft and Apple have also been reevaluating how much user data their on-device AI can access, often moving toward hybrid models that send anonymized data to servers for model improvements.

Why It Matters for Your Privacy

The original promise mattered because on-device processing meant you could use AI features without worrying about Google collecting behavioral data, search queries, or content from web pages. Without that guarantee, it’s possible that data from your browsing activity could be used to train AI models or personalize services—even if you’re not logged into a Google account.

This is particularly concerning for sensitive information you might interact with in Chrome, such as medical websites, financial portals, or private messages. While Google says it anonymizes data before sending it to servers, anonymization techniques have been repeatedly shown to be reversible under certain conditions. The lack of a clear, enforceable commitment creates uncertainty.

How On-Device AI Could Now Access More Data

Chrome’s on-device AI features—like “Help Me Write” in text fields, smart tab grouping, and AI-powered search suggestions—currently run locally. The change in privacy language suggests that future versions might allow these tools to communicate with Google’s cloud. Even if you choose not to use these features, Chrome’s underlying code could still collect telemetry about their performance.

One risk is that Google could expand data collection without asking for explicit consent, as long as it updates its privacy policy in the background. This kind of quiet change is not new in the tech industry, but it undermines trust, especially for users who rely on Chrome’s promises when making privacy decisions.

What Readers Can Do

You don’t have to accept this change passively. Here are practical steps to protect your data:

  1. Disable AI features in Chrome – Go to chrome://settings > “Advanced” > “Privacy and security” > “AI and machine learning”. Turn off any options that allow AI to access data. Keep in mind that this may disable some convenience tools.

  2. Review Chrome’s sync and account settings – If you’re signed into Chrome, check what data is being synced. Go to chrome://settings/syncSetup and disable everything you don’t absolutely need, especially history, passwords, and payment info.

  3. Use a privacy-focused browser – If Chrome’s direction concerns you, consider alternatives like Firefox, Brave, or DuckDuckGo’s browser. These browsers offer similar AI features but with clearer commitments to local processing or even no AI at all.

  4. Regularly check privacy policies – Companies often update their terms quietly. Set a reminder every few months to review Chrome’s privacy documentation. If the language changes again, you’ll be aware.

  5. Limit third-party cookies and tracking – Even without AI, Chrome tracks browsing activity for ad targeting. Use Chrome’s “Incognito” mode wisely, or install extensions like uBlock Origin and Privacy Badger for extra layers of protection.

What Experts Say

Privacy researchers interviewed by Decrypt expressed concern that Google’s decision creates a slippery slope. If the company can remove a privacy promise without notice, then similar guarantees for other features may also be vulnerable. “This isn’t a bug or a technical oversight—it’s a policy change,” one analyst noted. “Users need to assume that anything they do in Chrome could eventually be used for training unless they explicitly opt out.”

Sources

  • Decrypt, “Chrome Deleted Its Own Privacy Promise for Sneaky On-Device AI”, May 7, 2026. [Link]
  • Google Chrome Help (archived version via Wayback Machine for comparison).
  • Mozilla Foundation, recent statements on browser AI privacy.

The bottom line: Chrome’s privacy stance has shifted quietly but significantly. Whether you continue using it depends on how much trust you place in Google’s future data practices. In the meantime, adjusting your settings—or switching browsers—gives you more control.