Meta is planning to monitor US employees’ mouse movements, clicks, keystrokes, and screen activity. The goal is to use this data to train smarter AI agents that can handle workplace tasks. This move gives a glimpse of how far big tech companies are willing to go to develop automated systems that mimic human work behavior.

How Meta Will Collect the Data

Meta will use a tool called the Model Capability Initiative, or MCI, to gather information. The tool runs on work-related apps and websites and takes periodic snapshots of employee screens. According to internal memos, this data will help improve AI models, especially in areas where AI struggles to imitate human actions, such as navigating menus or using keyboard shortcuts.

Meta assured staff that the collected data will only be used for AI training purposes. It will not influence performance reviews. The company emphasizes that this initiative is aimed at making AI agents better at understanding and performing tasks in work environments.

Implications for Workplace Surveillance and Data Privacy

This move is likely to spark debates about worker privacy and data governance. Monitoring screen activity and keystrokes raises concerns about employee surveillance, especially as companies look to integrate AI more deeply into daily work routines. Critics worry about potential misuse or overreach, even if the data is meant solely for AI training.

In a separate memo, Meta’s CTO Andrew Bosworth explained that the company is expanding its internal data collection as part of a broader effort called the “AI for Work” initiative, now rebranded as the Agent Transformation Accelerator. Bosworth said the aim is for AI agents to eventually handle most of the work, with employees overseeing, reviewing, and helping improve these AI systems.

This shift reflects the broader trend in the tech industry. Companies like Anthropic and OpenAI have also showcased AI agents capable of managing software on behalf of users. Meta’s focus on capturing real employee behavior suggests a move toward AI that learns from actual workflows, not just predefined tasks.

What Experts Say About This Development

Experts believe this signals a shift from automating simple tasks to replicating entire workflows. Pareekh Jain, CEO of Pareekh Consulting, says AI systems will increasingly act as data pipelines, capturing how work is actually done so AI agents can perform it more effectively.

Sanchit Vir Gogia, chief analyst at Greyhound Research, adds that this isn’t just about documenting workflows anymore. Companies now want to understand how employees adapt and intervene during work. This real-time learning from employee behavior could make AI systems more flexible and capable but also raises privacy concerns.

Governance and compliance issues are also top of mind. Jain points out that capturing keystrokes and screen activity may violate privacy laws, especially in regions with strict regulations like Europe. Sensitive data such as credentials, intellectual property, or confidential workflows could be at risk, making security a major concern.

Gogia emphasizes that these risks don’t stand alone — they stack and interact. Balancing AI innovation with employee privacy and data security will be a key challenge for companies moving forward.

Inspired by

• https://www.computerworld.com/article/4161929/meta-to-track-employee-keystrokes-screen-activity-to-train-ai-agents.html

Sources

• reuters.com
• pareekh.com
• greyhoundresearch.com