‘This Makes Me Super Uncomfortable’: Meta’s Plan to Track Employees’ Every Click and Keystroke Sparks Backlash

Meta is rolling out a sweeping new program that turns everyday employee computer activity into AI training data — and it has already been met with internal backlash. 

Meta is installing tracking software on U.S.-based employees’ work computers to record how they use common work apps throughout the day, per Reuters. According to internal memos, the tool, called Model Capability Initiative (MCI), logs mouse movements, click locations and keystrokes. It also monitors how workers navigate software, including keyboard shortcuts and dropdown menus, and takes periodic screenshots or “screen content” to give the AI context for what’s happening on the screen.

Meta is rolling out MCI to U.S.-based full‑time employees and contingent workers. Staff will see a pop‑up on their work laptop asking them to enable the tool, which then runs on a pre‑approved list of work‑related apps and websites such as Gmail, Google Chat, Metamate (Meta’s internal AI assistant) and development tools like VS Code.

Meta will limit MCI monitoring to company devices and not track employees’ phones, the company said. It also stressed that it collects data “solely” to train AI models, not for performance evaluation.

Why Meta is doing this now

The program plays into Meta’s broader push to build powerful AI agents that can autonomously perform office tasks. The company recently created Meta Superintelligence Labs, a team of researchers and engineers dedicated to developing AI, as part of an overall AI effort. 

In memos viewed by Reuters, Meta stated that current AI systems remain clumsy at some of the mundane but essential behaviors of knowledge work. This includes activities like selecting the right option from a dropdown menu, using keyboard shortcuts efficiently and moving between multiple windows and apps to complete a task. 

To fix that, executives said the models need “real examples” of how humans actually use computers in the flow of work. By turning the daily activity of tens of thousands of employees into training data, Meta hopes to sharpen its internal AI agents so they can eventually perform end‑to‑end tasks on behalf of workers, per Reuters. The company also wants to gain an edge in the race with rivals like OpenAI, Google and Microsoft, all of whom are trying to create AI assistants for workplace tasks.

How employees responded

Inside Meta, the announcement has triggered visible discomfort and skepticism, as internal reaction threads obtained by Business Insider show. On Meta’s internal communications platform, the top‑rated comment under the rollout post read: “This makes me super uncomfortable. How do we opt out?”

Meta CTO Andrew Bosworth responded that “there is no option to opt out of this on your work provided laptop,” a message that drew crying, shocked and angry emoji reactions.

Meta told Business Insider that it will have safeguards to protect sensitive content and not use the data for any other purpose besides training AI. A source also noted that the company has monitored employee activity on company machines in some capacity for years.