👉 Background: HarperCollins is the British American publishing company that’s been around in some form since 1817. It’s the the publisher behind classics like The Lord of the Rings and The Alchemist, and has grown to become the second largest consumer book publisher in the world.

👉 What happened: So far, HarperCollins and other big publishers have been unwilling to engage with AI companies. But now, HarperCollins has become the first member of the Big 5 to invite AI into their library and has reportedly signed a deal with Microsoft to help train its AI models.

👉 What else: The deal will reportedly give Microsoft access and permission to HaperCollins' nonfiction catalogue to train its models. The compensation would be split between the publisher and the author as shared revenue and royalty streams.

What's the key learning?

💡When in doubt, hedge your bets—and your intellectual property. Many AI companies have been training their models on copyrighted material without asking permission (or paying for it).

💡 Training their models without permission has landed OpenAI and Meta in hot water with everyone from publishers to comedians. Think: NYT or comedian Sarah Silverman suing OpenAI and Meta. As a result, there are many legal cases on foot.

💡But nobody knows how these legal cases will land and what precedent will be set. So many publishers, media publications and authors are hedging their bets... just in case the judges decision goes against them. Recently, we've seen Open AI sign licensing deals with the like of NewsCorp, Reddit and more than 30 other publishers.