Sell or license company data for AI training

Does your company hold data AI labs want?

AI labs now license records of real work — support tickets with how they were solved, code with its review history, specialist documents — to teach models how a job is actually done. If your company has that kind of archive, tell us about it. We will tell you honestly whether buyers are asking for it, and who.

What buyers are asking for

Generic text that looks like the public web is worth little. What is in demand is work with the outcome attached:

Support and operations records that show the problem and the fix
Code history — commits, reviews, bug reports and how they were closed
Linked internal records — email, chat and tickets that refer to each other
Specialist documents in law, finance, medicine and engineering
Audio and video, especially footage that has never been published

What deals have been reported

These are public reports, not offers, and not a guide to what any one company will be paid:

Reddit licensed its content to Google for a reported $60 million a year (Reuters, 2024).
News Corp's deal with OpenAI was reported at over $250 million across five years (Wall Street Journal, 2024).
HarperCollins offered authors $5,000 per book for AI training, split with the publisher (reported 2024).
Archives from small companies that have shut down have sold for roughly $10,000 to $100,000 each (Forbes, reporting on SimpleClosure, 2026).

Check these before you go any further

Do you own it? Customer contracts often forbid reusing their data. Content you licensed from someone else is not yours to sell.
What did you promise? Data is bound by the privacy policy in force when it was collected. Quietly changing that policy later to allow AI training has drawn warnings from the US Federal Trade Commission.
Whose names are in it? Email and chat archives are full of employees' and outsiders' personal details. Removing names does not by itself make data anonymous.
Health records need proper de-identification under HIPAA before anyone else can see them.
Other strings: legal privilege, partners' trade secrets, earlier exclusive licences, and approval from investors or an acquirer.

None of this is legal advice. Talk to your own lawyer before any data leaves your systems.

Tell us about your data

Common questions

Will you buy our data?

No. GetAIGigs does not buy, hold or resell data. We tell you honestly whether what you have is the kind buyers are asking for, and introduce you to buyers who fit.

How much is company data worth?

There is no price list. Reported deals range from tens of thousands of dollars for a small company archive to tens of millions for the largest publishers, and most of the value depends on whether you actually hold the rights to sell. Anyone quoting a figure before seeing your data is guessing.

Do you charge for this?

No. If we are ever paid an introduction fee by a buyer, we will tell you before making the introduction.

Can we sell data that contains customer information?

Often not, or not without work first. Customer contracts and privacy policies frequently forbid reuse, and removing names does not by itself make data anonymous under laws like GDPR. Talk to your lawyer before anything leaves your systems.

I'm Anish T.

Software engineer · I build and maintain this site

I'm a backend and data engineer. I started taking AI-training contracts myself because the work fit around a full-time job — then spent months working out which platforms were real and which paid what.

There wasn't one place that just told you, so I built it. If you apply through Mercor and it asks who referred you, it's my name you'll see: Anish T. (GetAIGigs).

IndependentUpdated dailyNo estimated pay

How I make money, plainly:

Other than my goal of helping people find this work, the referrals make me money. If you sign up through one of my links and get hired, the platform pays me a fee out of its own pocket. It doesn't come out of your pay, it doesn't change your rate, and it doesn't change your odds of getting in.