A thread on Hacker News is exploding right now — over 1600 points and nearly 1100 comments — about something many Chrome users never knew: Google has quietly installed Gemini Nano, its on-device AI model, on your devices. Without asking. Without warning. Just dumped 4 GB into a folder called OptGuideOnDeviceModel in your Chrome profile.

The source is an article from the blog thatprivacyguy.com, and while this isn't an academic publication, the findings strongly resonate with people in the tech underground who have actually checked their drives and found the files there.

What makes this extra juicy is the re-download mechanism: if you manually deleted the model, Chrome automatically downloaded it again the next time you launched the browser. Not exactly what one associates with respect for user choice.

One researcher estimated that if the model were pushed to just one billion Chrome users, the distribution alone would consume 240 gigawatt-hours of energy and generate 60,000 tons of CO₂ equivalents.

Google confirms that from February 2026, they began rolling out a Chrome setting that allows you to disable and remove the model — but this came long after the download had quietly occurred for hundreds of millions of devices.

What is the model used for? According to Google itself: fraud detection, writing assistance («Help me write»), and developer APIs. All on-device, meaning your data is not sent to the cloud. That sounds good — but the problem isn't what the model does, it's how Google chose to distribute it.

The HN comment section is divided, as always. Some believe this is a legitimate technical decision since the model actually protects privacy by processing locally. Others are furious and draw parallels to previous cases where large platforms have used 'product improvements' as a pretext to take resources from users without consent.

Why should you care? Because this potentially affects everyone who uses Chrome — and that's about 3.5 billion people. If this starts to roll out in mainstream media in the coming days, it could become a real PR storm for Google amidst a period where trust in Big Tech is already under pressure. Additionally, it sets an uncomfortable precedent: if AI models can be quietly pushed to your devices via the browser, what's really the next step?

Check your Chrome settings under Privacy and security if you want to see if you can turn it off.