AI Companies Really, Really Want to Buy Your Work Files and Emails

Data brokers are collecting all sorts of internal corporate files to help train AI models. Not everyone is thrilled.

By Steven Levy | Business | November 6, 2026

In 2026, the race to train ever more capable AI models has intensified, and with it, a less visible but equally fierce competition: the scramble for data. AI companies are not just scraping the public web; they are increasingly turning to data brokers to acquire vast troves of internal corporate files, emails, and other proprietary documents. These datasets, rich with real-world business communication, are seen as gold mines for improving AI models' understanding of professional contexts, workflows, and nuanced language.

Data brokers, sensing a lucrative market, have begun aggregating and selling access to these internal files. They source them through various means—some legitimate, such as partnerships with companies willing to license their data, and others more questionable, like purchasing from third-party vendors who may not have explicit consent from the original creators. The result is a booming gray market where the lines between proprietary business information and AI training fodder are increasingly blurred.

Not everyone is thrilled. Privacy advocates warn that employees whose emails and documents are swept into these datasets may have never consented, raising serious concerns about confidentiality and data protection. Corporations, too, are waking up to the risks: sensitive internal communications, trade secrets, and personal information could inadvertently end up in the hands of AI developers—and potentially leak into model outputs. Some companies are now scrambling to update their data governance policies and legal agreements to prevent unauthorized sharing.

Meanwhile, AI companies argue that access to such data is essential for building models that can truly assist with knowledge work. They claim that without diverse, real-world corporate data, AI assistants will remain limited in their ability to understand context, tone, and the complexities of professional communication. But as the demand for data grows, so does the tension between innovation and privacy. Regulators are beginning to take notice, and the coming year may see new rules aimed at clarifying who owns—and who can sell—the digital exhaust of the modern workplace.

For now, the market for your work files and emails is thriving, and the question of whether that's a fair trade or a profound violation of trust remains hotly contested.

via Wired AI