// browser tracking

Source: Ownersnotrenters

Google is storing gigabytes of language models locally on users' devices through Chrome, bypassing traditional server-side processing. Data stays on-device, but the company still benefits from behavioral signals and model training. This marks a shift from the cloud-first model where all user interaction flows back to Google's servers. The shift is less about genuine privacy protection and more about regulatory positioning: local processing creates plausible deniability around data collection while still enabling Google to optimize its products through on-device user behavior. For brands and advertisers, this means the traditional "personal data" handshake with Google is being replaced by inference data—what users ask, search for, and generate locally—which Google can ingest without explicit consent frameworks.

Source: Kottke

Browser fingerprinting—the practice of collecting device data like fonts, screen resolution, and plugins to create unique identifiers—works without cookies or explicit tracking, making it invisible to most users and largely unregulated. As websites increasingly rely on this technique to bypass privacy regulations and ad blockers, consumers face a data collection problem they can't see or easily control. The "Taken" site demonstrates the technical feasibility of this tracking, putting pressure on browser makers and regulators to either build stronger defaults or watch fingerprinting become the primary surveillance method on the web.