What they're not telling you: # privacy-alarm-over-automated-scanning.html" title="Google Photos Update Sparks Privacy Alarm Over Automated Scanning" style="color:#1a1a1a;text-decoration:underline;text-decoration-style:dotted;font-weight:500;">Google removes privacy assurances after stuffing devices with their AI model Google has quietly deleted explicit privacy promises from its Chrome browser settings after unilaterally installing a 4GB artificial intelligence model on user devices, a pattern that suggests the company anticipated legal vulnerability. The timing is significant. An earlier version of Chrome's Settings interface displayed the on-device AI toggle within the System block alongside a specific assurance: "Without sending your data to Google servers." This was not marketing language—it was a legal representation.
# THE TAKE Google just executed a bait-and-switch that would make a three-card monte hustler blush. They stripped privacy language from their terms while simultaneously embedding Gemini AI into every conceivable device—phones, tablets, speakers, watches. The play is transparent: local processing creates a plausible deniability narrative. "*Your data stays on device,*" they'll claim. Except the training pipeline still requires upstream data harvesting. Voice queries, image recognition, behavioral patterns—all flowing toward Mountain View servers for model refinement. What changed? They stopped pretending otherwise. The old privacy assurances were theater. Now they're simply admitting the economic model requires surveillance infrastructure. This isn't evolution. It's capitulation. Google eliminated the language they could no longer defend legally once the AI integration made data extraction architecturally inseparable from functionality. The device isn't the product anymore. Your attention *is*.
What the Documents Show
Users in the EU, UK, and other jurisdictions are explicitly entitled to rely on a vendor's stated data processing claims when deciding whether to permit processing activities. The presence of this guarantee created a binding obligation: if a user left the default toggle enabled based on Google's assurance that data would remain local, the company became legally bound by that promise. That assurance has now vanished. The sentence promising that the model runs locally without transmitting user data to Google's servers has been removed from current Chrome settings screens. The deletion occurred after Google's silent installation of Gemini Nano—the 4GB model—began rolling out to users without explicit consent.
Follow the Money
When pressed on this practice, Parisa Tabriz, Google's senior vice president overseeing Chrome, claimed users could simply opt out, a response critics characterized as gaslighting given that most users never discover the setting exists. The mainstream tech press has largely treated this as a minor feature rollout accompanied by a philosophical disagreement about consent frameworks. What's been underplayed is the evidentiary trail the deletion creates. Removing a privacy representation immediately after deploying technology that contradicts it suggests deliberate intent to eliminate documentation of broken commitments. This isn't a technical update or interface refinement—it's the erasure of a promise made to users in jurisdictions with enforceable privacy law. The broader implication cuts to the core of how technology companies manage legal exposure.
What Else We Know
Google's approach demonstrates a pattern: make aggressive architectural decisions (pre-installing AI models), issue narrow technical responses to criticism (claiming opt-out exists), then methodically eliminate paper trails that might substantiate user claims in future litigation. Users who relied on the original privacy assurance and left the toggle enabled now have devices running Gemini Nano with no documented guarantee about local processing. The company has effectively rewritten the terms retroactively while maintaining plausible deniability about consent. For ordinary people, this means privacy promises embedded in software interfaces are increasingly disposable. If a vendor's assurance becomes inconvenient after deployment, nothing prevents its deletion. Users cannot reliably point to contemporary representations to prove they were misled, because those representations can be unilaterally removed.
Primary Sources
- Source: Hacker News
- Category: Tech & Privacy
- Cross-reference independently — don't take our word for it.
Disclosure: NewsAnarchist aggregates from public records, API feeds (Federal Register, CourtListener, MuckRock, Hacker News), and independent media. AI-assisted synthesis. Always verify primary sources linked above.