Workplace Telemetry Trains AI Agents

Big tech employers are now instrumenting employee desktops — collecting mouse movements, clicks, keystrokes and occasional screenshots — and feeding that telemetry into models to train AI agents intended to automate office tasks. Firms frame this as improving product capability and not for performance review, but the data collection blurs lines between product development, employee monitoring, and personnel governance. — Normalizing collection of granular employee interaction data for model training creates privacy, consent, labor‑rights and security tradeoffs that require public debate and potential regulation.