EBSCO’s publisher agreement for LLM training

We’ve already said we will not allow our author’s works to be used to train LLMs, but I want to share something that shows a bit about what is happening in that space in some detail. I received this email from EBSCO, to the address we have on file with them as a publisher whose books are included in their eBook products:

Hi Rory,

I hope your New Year is off to a great start!

Generative AI companies are actively seeking high-quality content to train their language models, and EBSCO is uniquely positioned to help publishers benefit from this opportunity. With our extensive network and trusted relationships, we’ve developed a secure model that protects your content’s integrity while creating a new revenue stream.

Here’s what we offer:

• We negotiate directly with AI companies to maximize the value of your content.
• All interactions—contracts, invoicing, delivery, and revenue collection—are handled by EBSCO.
• Strict safeguards: content is used only for model training and answering questions, never displayed or reproduced.
• Publishers earn $150 per title selected by the LLM

This is a chance to monetize your content without risk to your brand or existing revenue streams. Would you like to review the draft amendment and explore the details?

Looking forward to your thoughts,

xx

I don’t think most book contracts contain a clause that gives permission for the publisher to authorize this kind of use of the work. I think publishers are signing an agreement with EBSCO to include authors’ books in an LLM when authors haven’t signed away that right.

If you’re interested in seeing the draft amendment referred to in the email, drop me a line. I don’t think they would like me sharing that information, but I never signed anything or acknowledged anything saying I can’t or won’t. You can contact me at rory@litwinbooks.com