#privacy

Google is testing ads in AI Mode for people who connect Gmail and Photos, reversing its ad-free stance. What its help page says, what is excluded, what to do.

Customer Match lists can now carry an unhashed IP address and interaction time. The rules, the regions excluded, and who gains from adding the columns.

Test whether a local AI workflow depends on cloud models, search, embeddings or downloads, and record what works offline without overstating privacy guarantees.

Perplexity's Hybrid Compute splits a task: planning and search in the cloud, private files and sensitive steps on a 24 GB Apple silicon Mac, using no credits.

Seventeen coding agents, six questions each, vendor documentation only. What the published terms say on retention and training, and which cells stayed open.

SpaceXAI open-sourced Grok Build's Rust harness under Apache 2.0 days after a privacy scandal. Why a public repo is not the same as a security audit.

Meta is retiring the off-Meta-activity opt-out, rolling out US-first in July 2026. The replacement limits data use, not collection. What advertisers expect.

A clean room matches and measures with partners without sharing raw PII. Why most mid-market brands don't need one below ~$1M media spend, and when they do.

LM Studio 0.4.16 ships the Locally iPhone app plus LM Link, bridging your desktop models to mobile over an encrypted Tailscale mesh with no cloud relay.

A decision guide for when to generate synthetic training and eval data versus collecting real data: distillation, bootstrapping, and model-collapse risk.