Google Chrome has been quietly pushing a four-gigabyte Gemini Nano model onto user devices without asking permission, according to an EU-based security researcher whose findings raise fresh questions about how AI companies distribute on-device models.
The researcher, computer scientist and lawyer Alexander Hanff, says he verified the install by reading macOS kernel file system logs and traced the payload to a file named weights.bin tucked inside a folder called opt guide on device model. He caught Google Chrome downloading a four gigabyte Gemini nano model onto the user device. There was no consent prompt, Jaeden Schafer said on the podcast, summarizing Hanff's discovery.
Hanff argues the silent install runs afoul of European law. According to Schafer, the researcher claims the behavior "violates the e-privacy directive, article five three, which prohibits storing data on a user's equipment without prior consent plus GDPR article five and 25." Those provisions govern when companies can place files on a user's machine and how they must minimize and disclose data processing by design.
Key facts
- 01Security researcher Alexander Hanff says Google Chrome silently installed a 4GB Gemini Nano model on user devices with no consent prompt.
- 02The model file, named weights.bin, sits in a folder called opt guide on device model, verified via macOS kernel file system logs.
- 03Hanff argues the install violates the EU e-Privacy Directive Article 5(3) and GDPR Articles 5 and 25.
- 04Chrome has more than a billion users, meaning the silent download could affect storage and bandwidth at global scale.
The scale of the deployment is what makes the complaint sting. Chrome has well over a billion users, and a four-gigabyte payload is not a trivial background update — it is a sizable language model dropped onto consumer hard drives, potentially over metered connections, with no visible disclosure in the browser's normal update flow.
Schafer flagged the precedent as the bigger concern. This isn't really super great to give four gigabytes onto, you know, a billion plus Chrome users computers without any sort of, you know, telling them that they could do this, he said, warning that other AI vendors could follow Google's lead and treat large model downloads as routine software maintenance.
On-device models like Gemini Nano are central to Google's strategy of running AI features locally for speed, privacy and cost reasons. But that pitch depends on users trusting how the weights get there in the first place. Distributing them as an opaque background fetch undercuts the privacy argument that on-device inference is supposed to make.
The size issue will only intensify. Nano is the small end of Google's lineup, and it is already four gigabytes; successor generations are expected to be larger as capabilities expand. Schafer noted that "these models, the nano one, even they're four gigs, these things just get bigger and bigger," implying that what looks like a one-off footprint today could become a recurring storage tax on user devices.
Google has not publicly responded to Hanff's allegations, and no formal regulatory action has been announced. But with EU regulators already scrutinizing how large platforms deploy AI features, a complaint from a researcher who is both a computer scientist and a lawyer is unlikely to be the last word on what Chrome is shipping in the background.
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.
