Employers Harvest Workflows to Train AI

Updated: 2026.05.13 1M ago 2 sources
Companies are beginning to instrument employees' screens, keystrokes, mouse movements and app interactions to create datasets that train internal AI models. This practice ties worker behaviour directly into product development, performance evaluation, and cost‑cutting decisions. — If adopted widely, corporate harvesting of employee workflows will reshape workplace privacy norms, bargaining power, automation risk, and who benefits from productivity gains.

Sources

Meta Employees Launch Protest Against Mouse-Tracking Tech At US Offices
BeauHD 2026.05.13 95% relevant
Reuters reports Meta installed mouse‑tracking and will capture clicks/keystrokes to train models — a direct, named example of employers collecting everyday workflow telemetry to build AI systems.
NYT: 'Meta's Embrace of AI Is Making Its Employees Miserable'
EditorDavid 2026.05.10 100% relevant
Meta’s announcement to track what employees type/click and to use that data to train AI models, plus token‑use dashboards and performance review linkages, is the concrete exemplar.
← Back to all ideas