HarperCollins allows authors to license their books before sharing it to train AI

In today’s times, there are authors who are suing companies like OpenAI and accusing them of copyright infringement for using their books to train AI models without consent. To avoid this controversial situation, publisher HarperCollins is asking for permission beforehand.

Upon entering a deal with an undisclosed AI tech company, HarperCollins has agreed to share some of their non-fiction titles to train AI models. Here’s how this deal may prevent books from being used for AI training without consent and aid authors to get compensated of their work.

What does the licensing include and does it impact the author?

Among the authors who aren’t impressed with the deal, author and comedian Daniel Kibblesmith has shared screenshots of conversations with his agent. The publisher has requested his 2017 children’s book named Santa’s Husband to be included in training an AI language learning model.

Based on the discussion seen in the screenshots, the deal between the publisher and artificial intelligence company has been mentioned. It also shows that the publisher is offering authors a non-negotiable sum of $2,500 per title for a license period of three years.

ALSO READ: 5 tips for writing great AI prompts

In an effort to acknowledge and compensate the original creator, HarperCollins has left the decision with the authors. In a nutshell, if the author agrees to the deal, their book will be shared “with certain protections concerning credit and limits of verbatim usage per AI response.”

With the right regulations in place, it is encouraging to know that the publisher has addressed the deal with authors and provided them with a choice to agree or pass up on the deal. It seems to be a step in the right direction. However, since the company decides which books to be included in this training exercise, not all authors will be asked for consent to share their work to train AI.