Meta's Model Capability Initiative drew employee backlash after an internal data exposure, Times of India reports
An internal Meta program that collected employee device interaction data provoked internal protests and an exposure of collected material, prompting Meta to suspend the initiative and say future participation would be opt‑in.
In this brief: 3 sections 2 min read
MCI (Model Capability Initiative) proposed collecting keystrokes, mouse movements and screen activity from employees.
Purpose: teach AI systems how humans perform multi‑step computer tasks.
Initiative originated in an elite AI lab (TBD) at Meta.
Employees protested with memes, flyers and a petition reportedly signed by >1,800 workers.
A late‑June exposure showed data collected by MCI was visible internally, fueling backlash.
Meta reportedly switched off MCI within hours and later said it would make the program opt‑in if reintroduced.
Reports indicate uncertainty whether MCI data was used to train Meta's latest models.
At least one manager said it had not been used; a CTO comment suggested the data proved useful.