# What is data exhaust?

> What is data exhaust? The trail of data left as a by-product of activity. Forbes used a nearby phrase for startup Slack and Jira archives. The evidence is thin.

By Matt Burns, Vaultfish. Canonical: https://vaultfish.com/answers/what-is-data-exhaust  
Published: 2026-08-10  
Updated: 2026-10-14

## The short answer

Data exhaust is data left as a by-product of online activity, behavior and transactions. Cookies, log files and clickstream records are examples. In AI deals, Forbes said labs treat startup Slack archives as premium training data. Evidence is thin, with no public price for log files alone.

The phrase comes from data analysis and now turns up in AI licensing talk. The two uses overlap only partly.

## What counts as data exhaust?

[Wikipedia's](https://en.wikipedia.org/wiki/Data_exhaust) example is a bank. The transaction amount and the parties are the primary data. Whether the transaction happened at a cash machine or a branch is secondary. That is the exhaust.

Medical devices are another example in the entry. Pacemakers, dialysis machines and surgical cameras generate exhaust data while they run. It is typically discarded once a procedure ends.

Per the same entry, companies combine and study exhaust to improve products, personalize content and spot trends.

## Is data exhaust the same as operational data?

They overlap. AI buyers use "operational data" to mean the record of work: chat, tickets, email and documents. Exhaust is the by-product of using a system.

A Slack message is work. The log of who opened which channel is exhaust.

Our page on [operational data](/answers/what-is-operational-data) covers the buyer's meaning.

## Do AI labs buy data exhaust?

The evidence is thin. [Forbes' report](https://www.forbes.com/sites/annatong/2026/04/16/ais-new-training-data-your-old-work-slacks-and-emails/), summarized in its own post on X, said labs treat startup Slack, Jira and email archives as premium training data.

[Gizmodo's](https://gizmodo.com/failed-companies-are-selling-old-slack-chats-and-email-archives-to-train-ai-2000747916) report names workplace data, including documents, workflows and internal communications. It reports payouts of $10,000 to $100,000 per company.

I found no published price or named buyer for log files or clickstream records alone.

## Is there a privacy issue?

Wikipedia cites Shoshana Zuboff (2015). She describes turning behavioral data into predictive products as surveillance capitalism. Exhaust about people can still be personal information. An owner should ask counsel before any license.

Our [rights check](/rights-check) lists the ownership questions. The [risk map](/risks) lists the privacy ones. Talk to a privacy or commercial lawyer in your state before you sign.

Figure: comparison of Primary data and exhaust in a bank transaction: Primary data (Amount and parties), Exhaust (emphasized: The by-product of a transaction, Cash machine or branch). Exhaust is emphasized because it is the by-product, such as whether a transaction happened at a cash machine rather than a branch.

_Vaultfish helps you run the process. We are not lawyers and this is not legal advice. If you need counsel who knows this area, we can point you to one._

## Sources

- [Wikipedia's](https://en.wikipedia.org/wiki/Data_exhaust)
- [Forbes' report](https://www.forbes.com/sites/annatong/2026/04/16/ais-new-training-data-your-old-work-slacks-and-emails/)
- [Gizmodo's](https://gizmodo.com/failed-companies-are-selling-old-slack-chats-and-email-archives-to-train-ai-2000747916)