Tech Giants Buy Corporate Chat and Email Data as Public Internet Data Runs Dry

AI developers such as OpenAI and Anthropic are reportedly turning their attention to internal corporate collaboration data as they face limits in gathering publicly available internet data for model training.

Citing about 20 sources, including startup founders and data industry insiders, tech media outlet The Information reported on the 13th (local time) that demand for enterprise data—such as internal messenger chats, emails, video conference recordings, and code change histories—has surged over the past few months.

Bobby Samuels, CEO of data brokerage startup Protege, stated that fueled by this growing demand, his company's gross transaction volume has surged from $30 million (approx. 42.5 billion won) last year to at least $100 million (approx. 142 billion won) this year.

This demand is driven by the booming development of AI agents that perform actual tasks, such as customer service and invoice verification.

This is because records of live human interactions—such as emails containing developers' problem-solving processes, discussions on financial performance by CFOs, and software demonstration videos—are essential for AI training.

Data buyers primarily target startups facing bankruptcy or acquisition.

In fact, Warmly, an AI agent startup recently acquired by HubSpot, reportedly received four separate offers of up to $300,000 to purchase business communication data, such as internal meeting minutes and emails, after signing the acquisition agreement, but turned them all down.

As data trading becomes more active, de-identification work to address privacy concerns has also emerged as a major challenge.

Regarding this, Shub Sinha, CEO of data processing firm Integral, explained, "We are going through a rigorous anonymization process to maintain data value while complying with privacy regulations."
※ Please note: This article was translated by AI and may contain errors.