Meta captures employee mouse movements to train AI models

Meta deployed software on US employees' work computers to record their every keystroke, mouse movement, click, and screen activity without offering an opt-out in order to train AI agents, raising questions about workplace privacy and surveillance, informed consent, and the normalisation of treating workers as raw data. What happened Meta started to install software to record U.S.-based employees’ mouse movements, clicks, keystrokes, and screen context so it can train AI agents to handle computer tasks more like humans do. Internally called the Model Capability Initiative (MCI), the software captures every keystroke, mouse movement, click, and periodic screenshots as employees work across hundreds of websites and apps, including Google, LinkedIn, GitHub, Slack, and even personal Gmail. There is no opt-out for employees using company-issued devices. European employees are exempt because GDPR won't allow it. However, the controversy intensified after Reuters reported that internal documentation indicated MCI would capture the contents of emails and direct messages sent to US employees, regardless of where the sender was located, raising questions about whether the programme could create compliance challenges under EU law. Why it happened Meta's stated rationale is straightforward: if AI is supposed to act like a human using a computer, it needs real examples of how people actually work. The broader goal is to build AI agents capable of performing white-collar tasks on their own - the exact software Meta is racing to ship amid competition from OpenAI and Anthropic. The legal gap in the United States was a key enabling factor. US law is woefully behind in an age when most office work is done on a computer, and 'consent' is virtually meaningless when jobs are on the line. Under US federal law, there is currently no statutory limit on workplace keystroke surveillance on employer-owned devices. What it means The shift raises questions about consent, transparency and governance for data collected from employees, even when anonymised. The programme was disclosed via an internal memo rather than through any formal consent process, and employees were given no meaningful choice about participation. For employees. Several employees expressed concern that enhanced tracking could enable micromanagement at a time of job uncertainty. Pandemic-era remote work tools normalised monitoring for many firms, but specific logging to train AI models represents a step-change in scope. Multiple Meta employees described MCI as "dystopian" to reporters. For policymakers. The case exposes a critical gap in US employment and data protection law. European privacy law says you can't record employees' keystrokes without consent. American privacy law says nothing at all. This is not just another workplace monitoring dispute; it is a test of how far companies will go to turn ordinary office work into training data for autonomous agents. System 🤖 Unknown Developer: Meta Country: USA Sector: Technology Purpose: Train AI models Technology: Agen tic AI; Generative AI Issue: Consent; Employment/labour; Nor malisation; Privacy/surveillance; Proportionality; Transparency Timeline ⏰ April 2026. Meta announces the tracking initiative to staff. Tech executives firmly state that installation is mandatory on all U.S. work laptops with zero opt-out provisions available. April 21, 2026. Reuters reveals Meta has begun deploying MCI tracking software. April 22, 2026. CNBC reports MCI extends to Google, LinkedIn, Wikipedia and other third-party platforms. May 2026. Employees discover the software is tied to security tools and logging extensive data (clipboards, URLs, sleep cycles). Staffers publish critical internal memos, coin the phrase "Employee Data Extraction Factory," and launch a petition gaining over 1,500 signatures. UK union United Tech and Allied Workers begins recruiting Meta employees in response to MCI. May 20, 2026. Meta announces 8,000 corporate layoffs to fund it

Date it happened

2026-04-01

This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.