Google Chrome has been quietly downloading a four-gigabyte Gemini Nano model onto user devices without asking permission, according to an EU-based security researcher whose findings could put the browser on a collision course with European privacy law.
The researcher, Alexander Hanff, is a computer scientist and lawyer based in the EU, and he says he verified the install by reading macOS kernel file system logs. The payload arrives as a file called weights.bin, tucked into a folder named opt guide on device model. No consent prompt appears before the four gigabytes lands on the machine.
On the AI Chat Daily podcast, host Jaeden Schafer walked through the discovery, saying Hanff "caught Google Chrome downloading a four gigabyte Gemini nano model onto the user device. There was no consent prompt." Schafer added that the browser "didn't ask you, it just downloads this four gigabyte model onto your computer."
Key facts
- 01EU-based security researcher Alexander Hanff says Chrome silently installed a 4GB Gemini Nano model on user devices with no consent prompt.
- 02The payload is a file called weights.bin, stored in a folder named opt guide on device model, verified via macOS kernel file system logs.
- 03Hanff argues the install violates ePrivacy Directive Article 5(3) plus GDPR Articles 5 and 25, which require prior consent before storing data on user equipment.
- 04Chrome's user base exceeds one billion, meaning the same automatic download could land on a billion-plus machines if rolled out broadly.
Hanff argues the behaviour breaches the ePrivacy Directive's Article 5(3), which prohibits storing data on a user's equipment without prior consent, alongside GDPR Articles 5 and 25 covering data minimisation and privacy by design. Those provisions are the same legal hooks that have been used to challenge cookie banners and tracking pixels — and they treat any storage on a user's device, not just personal data, as a regulated act.
“he caught Google Chrome downloading a four gigabyte Gemini nano model onto the user device. There was no consent prompt.”— Jaeden Schafer
The scale is what makes the case unusual. Chrome has more than a billion users, and an automatic four-gigabyte push to that footprint is a different category of install from a routine browser patch. Schafer noted that pushing four gigabytes "onto, you know, a billion plus Chrome users computers without any sort of, you know, telling them that they could do this" is not really super great.
There is also a precedent problem looming behind the immediate complaint. On-device models are getting larger, not smaller, and Gemini Nano sits at the small end of the range Google ships. If browsers and operating systems can quietly stage multi-gigabyte model weights as part of background updates, users lose the ability to budget their own disk, bandwidth and compute for AI features they may never have asked to run.
Schafer flagged that risk directly on the show, warning that "every AI company could start downloading models onto your computer and say, it's just, you know, part of their software upgrade" and noting that "these models, the nano one, even they're four gigs, these things just get bigger and bigger."
Google has not yet publicly addressed Hanff's claims, and the researcher's analysis has not been tested in court or by a data protection authority. But the report lands at a sensitive moment for the company, which is racing to push Gemini features into Chrome, Android and Workspace while European regulators sharpen their focus on how AI systems collect, store and process user data. Whether Brussels treats a silent four-gigabyte model drop as a software update or as an unconsented install on user equipment may shape how every other AI vendor ships local models next.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
