HarperCollins seeks AI training rights from authors

The publishing industry continues to grapple with artificial intelligence as HarperCollins explores a new frontier in AI training data licensing, offering authors compensation for the use of their works in machine learning.

Key development: HarperCollins has established an agreement with an unnamed AI company to utilize select nonfiction books for AI model training, with authors having full control over participation.

• The publishing house is offering authors $2,500 for a three-year license to use their books in AI training
• The agreement currently focuses specifically on nonfiction titles from the publisher’s backlist
• Authors must explicitly opt-in to participate in the program

Industry context: This move comes amid ongoing legal battles between authors and AI companies over unauthorized use of copyrighted materials for AI training.

• Several authors are currently pursuing lawsuits against companies like OpenAI for alleged copyright infringement
• The opt-in nature of HarperCollins’ program directly addresses concerns about unauthorized use of authors’ works
• This represents one of the first formal attempts by a major publisher to create a structured licensing program for AI training

Publisher’s perspective: HarperCollins frames the initiative as part of their commitment to innovation while protecting authors’ interests.

• The company emphasizes that the agreement includes “clear guardrails around model output” to protect authors’ rights
• The program aims to balance new technological opportunities with the protection of existing revenue streams
• HarperCollins positions itself as an intermediary, presenting opportunities to authors while safeguarding their intellectual property

Looking ahead: This licensing model could set a precedent for how the publishing industry addresses AI training data rights, though questions remain about appropriate compensation levels and the long-term implications for authors’ creative control.