AI Data Deal TrackerDeal brief
HarperCollins offers authors $5,000 per title in AI training deal reported to be with Microsoft
HarperCollins agreed to license select backlist nonfiction books to an AI company for three years of model training. Authors must opt in and the reported fee is $5,000 per title, split evenly with the publisher.
NEW YORK, November 19, 2024. HarperCollins confirmed a licensing agreement with an AI company that allows limited use of select nonfiction backlist titles for training AI models. Bloomberg reported that the buyer is Microsoft, and press reports put the fee at $5,000 per book.12
Terms of the deal
The agreement runs three years and covers previously published nonfiction. Fiction is excluded. The reported payment is $5,000 per title, split evenly between the author and HarperCollins, and the money is not counted against existing advances.2 Authors must opt in, and Microsoft chooses which books it takes.2
Microsoft may reportedly use no more than 200 consecutive words or five percent of a book's text, and it pledged not to scrape text from piracy sites. The model is not expected to be used to generate books, and its purpose has not been announced.2
Background
HarperCollins was the first of the large US trade publishers to disclose an AI training license. The Bookseller reported on November 19, 2024 that the company had confirmed the agreement and that the information came from an email an author shared.1 The company did not disclose the payment or the buyer in that confirmation.1
Agents and authors reacted with mixed views. Some took the offer, while others questioned whether the price reflected the value of a book, and whether a three-year grant could be undone later.2
What it means for data sellers
The deal shows a workable template for rights owned by many individuals: a flat per-item fee, an opt-in, a short term and a cap on how much text the buyer can reproduce. The figure comes from press reports and was not confirmed by the companies. Sellers with large content libraries can use it as a reference point, remembering that nonfiction backlist and a company's internal documents are not the same asset.
How the deal came together
- An email to authors surfaced the deal. In November 2024 an author shared an email in which HarperCollins confirmed a licensing agreement with an AI company for select nonfiction backlist titles.
- Bloomberg identified the buyer. Bloomberg reported that the AI company is Microsoft. HarperCollins did not name the buyer.
- Authors were asked to opt in. Only authors who agree take part. Those who decline have their books excluded and are not paid.
- Safeguards were written in. The terms reportedly limit use to no more than 200 consecutive words or five percent of a book's text, and bar use of pirated sources.
Takeaways for companies with data
- A per-title price with an author opt-in gives buyers clean rights and gives sellers a defensible process.
- Payments outside the royalty account avoid disputes over earn-out of advances.
- Short licenses, here three years, keep the option to renegotiate as prices change.
Sources
Every numbered reference above points to an entry here. Figures come from these pages as of the date shown at the top of the article. Where a source reports a number it got from unnamed people, the text says so.
- HarperCollins signs contract with tech company to use limited number of titles to train AI (The Bookseller, November 2024)
- HarperCollins Book Catalog To Train Microsoft AI Models For The Next 3 Years (eWeek, December 2024)
###
About the AI Data Deal Tracker
SellYourAIData tracks publicly reported sales and licenses of data to AI companies, with the price, terms and sources for each. Entries are checked against public reporting and updated as deals change. SellYourAIData was not involved in this deal. To ask what your company's data could sell for, get a free valuation. To report a deal, contact us.