4GB

Check your Chrome user data folder. There’s a decent chance a 4GB AI model is sitting there—one you never agreed to install. The file is called weights.bin, buried in a folder named OptGuideOnDeviceModel. It’s the weight file for Gemini Nano, Google’s on-device language model.

Delete it and Chrome downloads it again.

Privacy researcher Alexander Hanff uncovered the behavior while running an automated audit on a fresh Chrome profile. Using macOS kernel filesystem logs, he traced Chrome creating a temp directory, pulling down model components, and placing the finished file on disk. The whole process took roughly 15 minutes. No notification. No prompt. The profile had received zero human input at any point.

The same pattern has been confirmed on Windows 11, Apple Silicon Macs, and Ubuntu. Users who’ve been finding unexplained storage spikes for over a year now have a name for the culprit.

What it actually does

Gemini Nano powers Chrome’s on-device AI features: Things like “Help me write an email,” scam detection, smart paste, page summarization, and AI-assisted tab grouping. On Windows, the file lands at %LOCALAPPDATA%GoogleChromeUser DataOptGuideOnDeviceModelweights.bin. On Mac and Linux, it’s the equivalent Chrome profile directory.

Deleting the folder provides no permanent relief. Chrome restores it on the next restart unless you disable the feature—via chrome://flags, the On-device AI toggle in Settings > System, or on Windows, a registry edit setting OptimizationGuideModelDownloading to disabled.

Chrome recently added a prominent “AI Mode” pill in the address bar. A reasonable user seeing that button—with a 4GB local model already on their disk—would assume their queries stay on-device. They don’t. AI Mode routes every query to Google’s cloud servers. The local Gemini Nano model doesn’t power it at all.

You’re paying the storage and bandwidth cost for a feature you’re not actually using privately.

Is it legal or “legal”?

Hanff argues Google is violating EU privacy law. His case centers on Article 5(3) of the ePrivacy Directive—the same clause behind cookie consent banners—which requires “prior, freely-given, specific, informed, and unambiguous consent” before storing anything on a user’s device. He also cites GDPR Articles 5(1) and 25, covering transparency and privacy by design.

He also drew a direct line to a case he published two weeks earlier: Anthropic’s Claude Desktop silently pre-authorized browser automation across roughly three million user machines without explicit consent. It’s the same pattern, he argued, but at a much smaller scale.

However, Google has been sneaking Gemini Nano in Chrome for a while. People just didn’t notice. “To provide an enhanced browser experience, Chrome uses on-device AI models to help power web and browser features,” Google says in its Support Site. “Chrome may download on-device Generative AI models in the background, so features that rely on these on-device models stay ready for use. If you delete on-device AI models, only features that rely on them will be unavailable.”

“In February, we began rolling out the ability for users to easily turn off and remove the model directly in Chrome settings. Once disabled the model will no longer download or update.” the company told Android Authority.

The company noted the model auto-deletes if storage runs low. What Google didn’t address is why users weren’t asked first.

Google’s own Chrome developer documentation tells third-party developers it’s “best practice to alert the user to the time required to perform these downloads.” Google didn’t follow its own advice this time.