Microsoft is the mystery company licensing HarperCollins books for AI training

The rapidly evolving landscape of AI training data has revealed Microsoft as the mystery company partnering with HarperCollins Publishers in a significant content licensing deal.

Deal specifics: Microsoft has entered into a three-year agreement with HarperCollins to use select nonfiction books for AI model training purposes.

• The arrangement specifically covers “select nonfiction backlist titles” from HarperCollins’ catalog
• Authors must explicitly opt into the training program
• Individual authors are being offered $2,500 per book for the three-year licensing period

Author reactions: The compensation structure has sparked discussion among writers about the true value of their intellectual property in the age of AI.

• Author Daniel Kibblesmith publicly shared that he received a $2,500 per book offer
• When asked about an acceptable offer, Kibblesmith stated he would consider it “for a billion dollars” or an amount that would eliminate his need to work, highlighting concerns about AI’s impact on creative professions

Project parameters: While details about Microsoft’s planned AI model remain limited, early information suggests a focused approach to content utilization.

• Sources indicate Microsoft does not intend to use the licensed content to generate AI-written books
• The company has declined to comment on specific details about the project
• The model has not yet been publicly announced

Industry context: This agreement follows a broader pattern of major tech companies securing publishing partnerships for AI development.

• News Corp, HarperCollins’ parent company, previously established a similar arrangement with OpenAI
• The OpenAI deal encompasses content from major digital outlets including The Wall Street Journal, the New York Post, and The Daily Telegraph

Strategic implications: The selective focus on nonfiction content and opt-in structure suggests a careful approach to AI training data acquisition, while raising questions about the future relationship between traditional publishing and artificial intelligence technology.