Doug Kreuzkamp was on a Zoom call in August 2026 when media reports told him that Google had bought data from Springshot, the airline operations company he runs. He learned of the sale less than 48 hours after it happened, without being directly notified.
His experience sits inside a new trade in the digital exhaust of office life: emails, Slack exchanges, Teams chats, code and operational records that can show an artificial intelligence system how work is done. AI labs building business agents are looking for large collections of real workplace activity to train those systems.
Brokers are presenting companies with six-figure offers, sometimes worth up to $2 million, for anonymised workplace data. The offers can expire within 48 hours, putting companies under pressure to decide quickly.
The sums are rising alongside demand. Gartner estimates that annual spending on data for AI training will climb from less than $1 billion last year to nearly $6.5 billion in 2027.
Google’s purchase of Spirit Airlines’ data in August 2026 showed the scale available in a single transaction. Google bought 100 million emails, 500 million Teams messages and 176,000 employee records for $10 million.
Senator Elizabeth Warren and over 100 other members of Congress subsequently wrote to Google and Spirit, asking them to protect the privacy of thousands of former Spirit Airlines employees. The airline’s flight attendants’ union also objected to the sale on privacy grounds.
The concern is straightforward: staff may have created the messages and records, yet are often unaware that their internal communications can be sold on. The proceeds generally go to senior creditors and investors rather than employees.
SimpleClosure, which helps with startup liquidations, added data sales to its services in April 2026. Its Asset Hub has handled over 150 deals and expects to return over $10 million to startups by the end of 2026. Code or data sales now appear in 47% of SimpleClosure’s deals.
Keelan Miskell, founder of a crypto startup, received roughly $25,000 for the company’s code base. He also has a still-pending payment, likely around $15,000, for workspace data held in Google, Notion, Figma and Slack.
Dan Novaes, the chief executive of Mode.Inc, a startup that sells data to AI model developers, described the appetite bluntly: 'Right now, the demand is just astronomical'.
Large technology companies are also looking inward. Meta announced a programme earlier this year to track employees’ keyboard inputs and mouse movements for AI training. Andrew Bosworth, Meta’s chief technology officer, said the tracked data would be 'strictly controlled'.
Employees pushed back, and an internal leak revealed that data presented as anonymised included private staff conversations and performance records. Meta suspended the programme after the leak and did not restart it.
At a SpaceX all-hands meeting in August, Elon Musk told employees that the company would use employee data to train its Grok AI. Bloomberg reported in September that SpaceX’s AI unit was weighing purchases of data from struggling startups.
The legal guardrails remain thin. There is no federal law banning the sale of employees’ workplace data. Greg Ewing, an attorney who advises AI startups, has called such sales 'legally radioactive'. Ifeoma Ajunwa, a law professor at Emory University, said: 'Employers should have express consent before selling workers’ data'.
By October 2026, reports had laid out a market in which data generated during ordinary work can be sold as raw material for AI systems. The price may flow to creditors and investors. The messages, decisions and records at issue were generated by workers who may never be told they were part of the deal.
Comments (0)