Canada's Privacy Commissioner found OpenAI violated Canadian privacy laws in developing the original ChatGPT, adding a formal regulatory finding to the AI training data debate.
Harvard/MIT: adversarial prompts make email-forwarding agents hand over SSNs and erase memory of the act. 'The more powerful the agent, the bigger the blast radius.' Urgent safety signal.
Emotion AI facial tracking tools are entering white-collar offices as 'wellbeing monitoring,' with no consent frameworks and documented bias concerns. The Atlantic reports on its rapid spread.
Bloomberg investigation: nearly 20 US state health exchanges transmit race and citizenship data to Meta, TikTok, and Google via ad pixels — classified as protected health information under HIPAA.
Gemma 4 E2B + WebGPU + Transformers.js enables a fully local Chrome browser agent with no server calls — tabs and browsing data stay on-device.
GitHub CLI now collects pseudonymous telemetry by default — no opt-in offered — affecting all users globally and requiring active opt-out to disable.
OpenAI's Privacy Filter — a 1.5B MoE model for PII detection under Apache 2.0 — is the company's first open-weight release of 2026, running on-device with 128k context.
Curated AI insights — sent when there's something worth your inbox.