Meta’s workplace surveillance expands: AI training data extraction entrenches extractive labor practices under guise of productivity
Original framing: “Meta to capture U.S. employee mouse movements and keystrokes to train AI” — The Japan Times
The extraction of worker productivity data mirrors 19th-century company towns, where employers controlled housing, wages, and even workers’ bodies to maximize output. Fordist assembly lines institutionalized time-motion studies to extract every possible unit of labor, foreshadowing today’s keystroke logging and screen monitoring. The gig economy’s algorithmic management continues this tradition, but with digital tools that obscure the power dynamics. This historical continuity shows how surveillance is not a bug but a feature of capitalist labor extraction.
Meta’s expansion of workplace surveillance is not an isolated technical decision but a symptom of a broader extractive logic that treats human cognition as a raw material for AI training.
This practice is rooted in the historical continuity of labor control, from Fordist time-motion studies to today’s algorithmic management, while being justified by Silicon Valley’s myth of meritocratic innovation. The framing obscures the racialized and gendered hierarchies of tech labor, where marginalized workers bear the brunt of surveillance, and ignores Indigenous and Global South critiques of data commodification. Without systemic interventions—such as worker co-ops, public digital infrastructure, and global labor standards—this model will entrench corporate power, deepen inequality, and erode the last vestiges of workplace autonomy. The solution lies in reimagining data not as a corporate asset but as a collective right, governed by democratic principles rather than shareholder greed.