Companies Are Selling Employees' Private Slack Messages and Emails to Train AI—Often Without Workers' Knowledge

Companies are increasingly licensing workplace emails, Slack messages, and internal documents to AI developers, creating a lucrative market that some employees may not know about. As AI firms race to automate office jobs, businesses are receiving six-figure offers for their digital records, sometimes with just 48 hours to decide.

These deals can generate substantial revenue, but they also raise questions about employee privacy, consent, and who should benefit financially from the information workers create. 'Right now, the demand is just astronomical,' said Dan Novaes, chief executive of data broker Mode.Inc., speaking to Business Insider.

Business Data Has Become the New Oil

AI developers, including OpenAI, Anthropic, and Meta, are seeking real-world workplace records to train AI agents capable of performing tasks traditionally handled by employees. Gartner's September 2026 forecast confirms that worldwide spending on AI data is projected to reach $6.48 billion in 2027, reflecting growing demand for data used in AI development.

Brokers approach businesses with extensive archives, offering to buy their records, remove sensitive information, and license the remaining data to AI developers. Offers can depend on the volume of information, its exclusivity, and the industry involved.

Ryan Lock, co-founder of AI data broker Polyshares, compared the practice to early oil prospectors who bought land from farmers unaware of the oil beneath it. 'The offers being made look almost identical to those early oil offers,' he said. 'They'll say, "I'm offering you double market. I need an answer by Friday."'

Employees, meanwhile, may receive no share of the proceeds, despite generating the emails, presentations, and internal conversations that make the data valuable. Ifeoma Ajunwa, a law professor at Emory University, described the question of whether employees' consent extends to the subsequent sale of their data as an 'ethically gray area'.

Spirit Airlines Data Deal Raises Employee Privacy Concerns

The risks extend beyond active businesses. Google's proposed $10 million purchase of internal Spirit Airlines data has raised concerns among former employees and lawmakers. The airline, which ceased operations in May 2026, had an archive containing approximately 100 million emails and 500 million Microsoft Teams messages.

On 8 October, more than 120 US lawmakers, led by Senator Elizabeth Warren and Representative Steven Horsford, urged Google and Spirit Airlines to protect former employees' privacy in connection with the proposed $10 million data deal. Google said it was not seeking personal information and that such data would be excluded or de-identified by an independent third party, according to reports.

The dispute highlights a difficult question: does an employee's consent to workplace data collection also extend to the subsequent sale or licensing of that data for AI training?

Data Access and Security Add to Privacy Fears

The risks are not limited to legitimate data sales. Nick Heiner, who works on AI training environments at Surge AI, told Business Insider that someone had approached him offering a large company's codebase for millions of dollars. He concluded the group did not have the rights to sell it.

Meta has also faced internal backlash over plans to collect employees' keystrokes and mouse movements for AI training. The initiative raised fears that sensitive information could be captured. According to reports, data intended to be anonymised became accessible more widely within Meta than intended. The material reportedly included employees' private conversations and performance records. Meta suspended the project following the incident.

A Growing Market With Unsettled Rules

The growing use of employee records to train AI could intensify calls for stronger privacy protections, said Alex Rosenblat, a sociologist and former Uber policy executive. It could also give employees new leverage and negotiating power as the value of their everyday digital footprint grows, she said.

'It's not that data collection wasn't present before; it's that the scale at which it can happen, the scale at which it is useful, has shifted,' Rosenblat said. 'And that's an opportunity for people to say, "wait a minute, the value of this is escaping us."'