Hilco and selling AI training data in a distressed sale

Hilco works with companies in distress, their lenders and the courts. It is not a program you apply to as a healthy business.

There is no price list. Hilco runs sale processes where bidders compete, which is how distressed data has drawn AI buyers: in the Spirit Airlines case, Google bid $10M and Mercor $7.5M, and micro1 later filed a $12.5M bid.

A live Hilco listing bundles source code and technical files with 53,712 engineering emails written between 2017 and 2024. AI data buyers look for precisely this sort of archive.

Terms depend on the sale, the court and the buyer. Hilco's own fee is set by its engagement; ask who pays it and how it is calculated before you sign.

Hilco publishes no referral program for introducers.

Hilco fits a company in or approaching bankruptcy where lenders or a court will run the sale of assets. If you are closing on your own terms, a direct buyer or a wind-down service is usually simpler.

We run the same checks on every buyer we rank. They take an afternoon and they protect the one thing you cannot take back: the copy of your records.

• Find the legal entity. A registered company name, an address and at least one named person. A Facebook form alone is not enough.
• Get the offer and the contract in writing before any export. A sample agreement tells you more than a sales call.
• Ask who strips personal data, how, and when. You should approve the method and see a sample before anything leaves your systems.
• Ask where the data goes. Which AI companies receive it, whether you are told, and whether it can be resold.
• Ask how everyone is paid. By you, by the buyer, or out of your deal. Referral fees are normal; hidden ones are not.
• Get a second quote. Terms from at least one other buyer is the fastest way to know if an offer is fair.