Tech workers push back against AI training that spies on workers

Company insiders say they’re fighting a new system that watches every mouse twitch and keystroke to teach artificial intelligence. Teams in multiple offices are posting flyers that ask blunt questions: why should daily work turn into free training material for company AI? The anonymous handouts call the setup an “Employee Data Extraction Factory” and point out that privacy checks required for handling staff data aren’t even finished. Employees who signed an online protest say the company’s promises about protecting their information feel hollow. Bosses offered executives an easy way to skip the tracking, yet most workers have no opt-out at all. The mood inside offices has shifted from routine coding to awkward questions: if the computers are learning from real humans, how much of their private workflow are we giving away without real consent?Last spring, company leaders announced they would start filming screens and logging clicks to make AI behave more like people. A research scientist told staff the goal is fixing areas where the software still acts like a clueless robot. In plain words, your everyday spreadsheets and Slack messages are now training material. Internal memos reveal the company plans to collect even more raw data under a sweeping AI spending plan worth over a hundred billion dollars. A top tech officer described a future where most tasks are handled by automated agents while humans mostly supervise. Exactly how these agents will learn wasn’t spelled out, only that the company would “build up data and evaluations” from every interaction at work. Spokespersons defended the move, saying AI needs to see real examples of how people move and click to improve. Without this constant stream of ordinary actions, they claim the models won’t get better at everyday tasks. Critics wonder whether workers’ privacy is the hidden cost of cheaper automation that can eventually replace many roles.