Meta has begun monitoring employee computer activities to train AI models, sparking discussions about workplace surveillance and automation.
Meta is rolling out software across its U.S. workforce to track employee activities, including mouse movements, clicks, and keystrokes. This data will be utilized for training artificial intelligence models, as reported by internal documents.
The initiative, known as the Model Capability Initiative (MCI), aims to enhance AI agents' ability to autonomously perform job-related tasks. The software is designed to operate within work applications and periodically capture screenshots to gather insights on user interactions.
One of the primary objectives is to improve AI performance in specific areas, such as navigating dropdown menus and utilizing keyboard shortcuts. According to the documents, “This is where all Meta employees can help our models improve simply by doing their everyday work.”
Andrew Bosworth, Meta's Chief Technology Officer, indicated in a memo that the company plans to ramp up data collection as part of its AI for Work initiative. He stated, “We aim for agents to do the work while you guide and review them, helping them improve.”
Meta spokesperson Andy Stone reassured that the data collected through MCI will be strictly for training purposes and will not impact employee performance evaluations. Measures are in place to secure sensitive information.
Impact on Employment
As AI tools increasingly automate tasks, companies are reducing personnel costs, with Meta planning to cut 10% of its workforce starting May 20 and considering further layoffs by year-end. Other firms, like Amazon, have recently let go of 30,000 workers, while Block reduced its staff by nearly half.
Despite potential short-term slowdowns, Meta promotes the use of AI agents for tasks such as programming. Recently, the company created a dedicated team focused on Applied AI to enhance neural network programming capabilities, eventually leading to the development of new products and infrastructure. Top programmers have been reassigned to this team.
Legal Considerations
Legal experts highlight that monitoring employee activities through keystroke tracking represents a significant escalation in workplace surveillance, typically reserved for gig workers. Yale Law School Professor Ifeoma Ajunwa noted, “There are no federal restrictions on employee monitoring in the U.S.” Conversely, York University law professor Valerio De Stefano pointed out that such practices would likely conflict with European laws.
In addition, Meta recently introduced the AI model Muse Spark, developed by its new Applied AI team, further emphasizing its commitment to AI advancements.
