AI companies are offering businesses up to $2 million for something their employees create every day

The demand for live, operational workplace data is intensifying as AI developers exhaust publicly available datasets. Companies like OpenAI, Anthropic, and Meta are reportedly working with data brokers to acquire specific "long-horizon" data that illustrates how work processes unfold, rather than just static information. This shift reflects a need for more dynamic and context-rich data to develop advanced AI agents capable of automating white-collar tasks.

The value of this data is influenced by exclusivity, a company's field, and the sheer volume of information, with prices based on terabytes. Businesses with more white-collar employees are seen as having more valuable data corpuses. The report highlights that this data market is distinct from traditional workplace surveillance, yet it introduces new dimensions of employee privacy concerns, particularly regarding the monetization of daily interactions without employee consent or benefit.

A key concern arises from the ethical implications, as current federal privacy laws in the US do not explicitly prevent companies from collecting and selling employee data for AI training. The question of whether initial employee consent for data collection extends to its sale remains an "ethically gray area." This situation creates potential for employee pushback, as demonstrated by Meta employees who successfully resisted a program to collect their keystrokes and mouse movements for AI training.