Meta Scales Back Tracking Tool After Employee Pushback, Adds 30-Minute Pause and Exemption Options
Miles Bennett
Meta is scaling back MCI, an internal tool that harvests employee computer activity to train AI models, adding a 30-minute pause and broader exemption options — a direct concession after more than 1,500 employees petitioned against the program, exposing a deepening tension between Big Tech's AI data appetite and workforce privacy.
What exactly does this tracking tool collect?
MCI (Model Capability Initiative) launched in April. It records mouse movements, button clicks, and drop-down menu navigation on employee computers.
The goal is to teach AI models to "use a computer like a human." In plain terms = Meta wants AI to learn software operations by watching employees work.
Zuckerberg's own words at an all-hands meeting: "AI models learn by watching very smart people do things."
Why are employees pushing back?
Concerns cluster around three issues: personal data exposure, battery drain, and no control over when data is collected.
Some employees resisted by repeatedly ignoring the permission pop-up's "Accept" button; others disabled the software through device settings.
Last month, employees handed out flyers at U.S. offices and launched an online petition demanding a halt to data collection. More than 1,500 people signed.
What concessions did Meta make?
Pause function: employees can pause data collection for up to 30 minutes; the tool resumes automatically when time expires — the memo says this lets staff "look at some personal things."
Wider exemptions: employees handling sensitive content, remote workers worried about bandwidth bills, and field workers who cannot keep laptops plugged in can all now apply for exemption.
This means → Meta has not cancelled MCI. It has offered employees a limited opt-out window while the core data-collection logic stays intact.
What does this standoff signal?
Superintelligence Labs VP Stephane Kasriel acknowledged employee concerns in the memo, yet stressed that existing privacy safeguards "went through multiple layers of risk review."
In plain terms = the company's stance is "we didn't do anything wrong, but we hear your frustration."
This reflects a larger collision: Big Tech's AI training increasingly depends on real behavioral data from its own workforce, and employees are drawing the line through collective action — a 30-minute pause is the starting point of a compromise, not the end.
Content is for reference only, not financial advice.