Meta Suspends Internal Employee AI Data Program

Meta halted its Model Capability Initiative (MCI), which mandated employees' keystrokes and mouse movements be recorded to train AI models to mimic human computer use, after over 1,800 employees signed a petition and a data leak exposed the program's lax security. Meta says the program remains suspended two months later and would only be reintroduced on an opt-in basis.

4 effects · 1 potentially affected company · The full cascade analysis is part of the Pro plan. Create free account

Meta loses a proprietary internal data source it hoped would give its AI unit an edge in building agentic, computer-using AI models, a capability central to automating white-collar work and competing with OpenAI and Google. The episode signals that employee resistance is becoming a real constraint on frontier AI labs' data-acquisition strategies, following similar friction at Spirit Airlines/Google over flight-attendant data. Other AI labs may face heightened internal and public scrutiny if they attempt similar workforce surveillance-for-training schemes, potentially slowing a key vector of proprietary data advantage. Meta's AI chief Alexandr Wang and CTO Andrew Bosworth face internal credibility costs, with Bosworth publicly blaming activists for making the company "worse." The affair could accelerate calls for opt-in-only data policies and labor protections across the AI industry, raising compliance costs for firms pursuing similar internal data harvesting.