Converging signals point to the integration of behavioral tracking at the very core of the workspace.
Since the early days of machine learning, AI has always been about observing and reproducing patterns—and that's fine in itself. But once this becomes a built-in system of the workplace, continuously capturing and structuring human activity, the slope turns ethically dangerous. In this particular case, it looks like a striking pivot—after years of leading open, frontier AI research—toward something far more extractive.
In a striking paradox, it could take here two equally dehumanizing forms: on one side, outsourcing low-value, residual cognition to low-wage, human annotation factories; on the other, capturing human employee behavior in real time—clicks, flows, micro-gestures—to be later replicated.
Together, they shape a true extracting pipeline.
This is not an AI that augments, but rather an AI that drains, copies, to later replace at scale.
And there is a deeper symmetry at play: if the customer is reduced to a graph of relationships to extract and monetize, then the employee ultimately becomes a stream of behaviors to extract.
Same logic. Same system. A vast—and quiet—parody of human work and relationships.
