Failed Startups Sell Old Chats to Train AI

Failed startups are now selling their internal data, like Slack chats and emails, to AI companies. This data is being used to train new AI models, fetching prices up to hundreds of thousands of dollars.

Digital Detritus of Failed Ventures Now A Commodity

Companies winding down operations are peddling their internal archives—think Slack chats, emails, and project logs—to artificial intelligence developers. These once-discarded digital fragments have become a hot commodity, fetching substantial sums, sometimes in the hundreds of thousands of dollars, to train sophisticated AI models.

A company named SimpleClosure is facilitating these transactions, offering a service called Asset Hub. This platform allows shuttered startups to monetize their data, which includes years of unfiltered workplace communications. AI firms, in turn, are actively bidding on these archives, often outmaneuvering traditional asset buyers. This trend underscores a shift in what constitutes valuable data for the burgeoning AI industry.

The Data Gold Rush

The demand stems from the evolving needs of AI development. While early models relied on public data like books and websites, newer, more advanced AI agents require datasets that mimic real-world work scenarios. Internal company communications, rich with nuanced human interaction and decision-making processes, are seen as crucial for training AI agents to perform complex workplace tasks.

Read More: Bridger WESTERN Macros Used in Online Game, Affecting Fair Play

AI companies are specifically seeking unfiltered conversations, detailing everything from technical debates to strategic decisions that may have contributed to a company's downfall. This data is considered superior to synthetic, AI-generated content, which can lead to detrimental feedback loops that degrade model quality.

Privacy Looms Large

The practice, however, is fraught with significant data privacy concerns. Even when attempts are made to anonymize the data, personally identifiable information can easily persist, especially for long-term employees. Many workers whose communications are being repurposed were reportedly unaware that their past conversations could become fodder for AI training. Privacy advocates have long warned about the potential for exploitation of digital workplace communications, highlighting that what feels private can readily become accessible data.

While ownership of company data often defaults to the company or its liquidators upon closure, employees typically have minimal control over how their historical communications are used.

Read More: Fifth Third Bank's Profit Drops After Comerica Deal in Q1 2026

The Broader Context: Data as a Battleground

This trend surfaces as data itself becomes an increasingly contested resource in the AI landscape. Earlier this year, Salesforce implemented policies to restrict rival AI firms from accessing Slack data, signaling a move to control this valuable asset. This restriction forced AI platforms reliant on Slack data to seek alternative sources, demonstrating that direct access to real-time communication is evolving into a competitive advantage. The struggle for data control illustrates how different entities are positioning themselves in an arena where data access is directly tied to enterprise value.

Frequently Asked Questions

Q: Why are failed startups selling their old digital records?
Companies that are closing down are selling their internal data, such as Slack chats and emails, to artificial intelligence developers. This data is used to train AI models.
Q: How much money can these companies make from selling data?
These digital archives can sell for significant amounts, sometimes reaching hundreds of thousands of dollars. AI firms are paying these prices to get data for training.
Q: What kind of data are AI companies buying?
AI companies are buying unfiltered internal communications, including Slack chats, emails, and project logs. This data is valuable because it shows real-world work scenarios and decision-making processes.
Q: Are there privacy risks with selling this data?
Yes, there are major privacy concerns. Even if companies try to remove personal information, it can still be present in the data. Many employees whose conversations are sold were not aware this could happen.
Q: Who controls this data when a company closes?
When a company closes, the data usually belongs to the company or its liquidators. Employees typically have little say in how their past communications are used after they leave or the company shuts down.