Meta's previously announced internal AI training program has encountered significant setbacks. The tool, named "Model Capabilities Initiative" (MCI), was originally designed to collect real computer usage data by tracking employees' keyboard keystrokes and mouse clicks to train its AI agents. However, this aggressive approach sparked major controversies and strong backlash within the company.
More than a thousand employees felt that being tracked for their daily behavior had a strong "dystopian" feel, especially against the backdrop of recent large-scale layoffs at the company. In response to a petition signed by over 1,500 employees, Meta eventually made compromises and reduced the scope of the program.
A Dual Compromise on Privacy and Technology
According to newly disclosed internal memos, Meta adjusted the rules to allow employees to apply to pause data collection once per day, but each time only for a maximum of 30 minutes. Additionally, a small number of employees working on sensitive tasks could be exempt from software monitoring upon approval from their superiors.
Aside from employee resistance, the reduction of the program was also accelerated by technical failures. Due to the massive amount of data required by the tracking tool, some remote workers experienced abnormal network traffic spikes, which even severely affected laptop battery life.
The Trust Dilemma in Enterprise AI Deployment
Meta executives acknowledged in internal memos the concerns raised by teams regarding data control and battery life, stating that multiple optimization measures have been introduced. Although executives emphasized their confidence in privacy protection measures, these technical fixes clearly cannot easily eliminate employees' deep-rooted suspicions.
This incident reflects the core contradiction faced by tech giants when deploying AI: the intense demand for real data for model training versus the boundaries of employees' personal privacy. As public data copyright disputes continue, turning attention to internal employee data is putting companies into a more serious internal trust crisis.
