Memorandum
- From
- Derek Weston via Fast Company
- Date
- Filed
- Business·4 min to read
- Re
AI Firms Offer Businesses Up to $2 Million for Workplace Data
ReAI Firms Offer Businesses Up to $2 Million for Workplace Data
AI data brokers are paying companies between $100,000 and $2 million for operational data — including Slack messages, CRM records, and email histories — to train AI agents, raising privacy concerns among employees and lawmakers.
AI companies racing to build autonomous agents are turning to an unexpected source for training data: the everyday digital exhaust of active businesses. Data brokers are offering companies between $100,000 and $2 million for their operational records — Slack messages, customer relationship management logs, email histories, and other internal communications — according to a new report from Business Insider.
The market has become highly competitive as firms like OpenAI, Anthropic, and Meta exhaust the data available from defunct companies and now seek live information from operating businesses. Micro1, a Palo Alto startup, is offering an unlimited $50,000 referral bonus for any company approved for its data partnership program. William Almeida, Micro1's chief revenue officer, told Business Insider that the most valuable data is «long-horizon» information showing how work actually gets done.
Pricing depends on several factors. Data is sold by the terabyte, and companies with more white-collar employees command higher offers. Exclusivity and a company's field of expertise also affect valuation. Dan Novaes, CEO of Mode.Inc, described the demand as «astronomical» and urged business owners to recognize the opportunity. Companies that refuse to sell their complete dataset — including all communications and records — find the remaining data is worth significantly less.
The trend represents a new form of workplace surveillance, and it is already generating pushback. Earlier this year, Meta employees revolted against a program that would have collected keystrokes and mouse movements to train AI models, forcing the company to suspend the effort. A Meta internal leak later revealed that supposedly anonymized data, including private conversations and performance records, was visible to unauthorized parties.
In a separate case, a longtime remote employee in New Jersey filed a lawsuit against her former employer after she refused webcam monitoring that would have photographed her every three minutes. The case highlights growing tension over how much control companies have over employee-generated data.
Google's $10 million bid for Spirit Airlines' entire dataset of internal communications, spreadsheets, calendars, transactions, and frequent flyer information during the airline's bankruptcy proceedings has also drawn congressional attention. Last week, 119 members of Congress — mostly Democrats — signed a letter co-written by Senator Elizabeth Warren urging Google and Spirit to protect the privacy of former Spirit employees.
Legal experts note that no federal privacy laws currently prevent companies from collecting employee data for AI training. Ifeoma Ajunwa, a law professor at Emory University, told Business Insider that the core question is whether employee consent extends to the sale of that data. «I think it's an ethically gray area,» she said.
Sensitive information such as Social Security numbers and home addresses is supposed to be scrubbed before sale, but the Meta leak demonstrates that anonymization does not always work as intended. Many employees remain unaware that their daily workplace interactions are being monetized, and they rarely share in the profits.
The emerging market for mundane workplace data could face growing scrutiny as workers, regulators, and lawmakers begin to question who truly owns the digital traces of a day's work. For now, the offers keep coming, and the sums are large enough that many business owners are listening.
6
