A panel discussion about artificial intelligence kicked off this year’s Frankfurt Book Fair’s virtual digital rights meeting, on September 2. It brought together Chantal Restivo-Alessi, chief digital officer and CEO international foreign language at HarperCollins; Dr. Jessica Sänger, director for European and international affairs at the German Publishers and Booksellers Association; and Fatimah Abbas, international publishing consultant and AI strategist at Cairo’s FALA Agency, to examine how artificial intelligence licensing deals are reshaping the publishing industry’s landscape.
Major publishers are pursuing licensing agreements with artificial intelligence companies while simultaneously implementing protective measures against unauthorized use of their content, as the industry navigates an uncertain regulatory landscape that could reshape publishing’s economic model.
Balancing copyright protection with potential revenue
The dual approach reflects the complex reality facing rights professionals, who must balance copyright protection with potential revenue opportunities from AI training deals. Three business models are emerging in the space: collective licensing similar to music industry practices, revenue-sharing arrangements based on AI usage, and strategic partnerships between publishers and technology companies.
“AI can be publishing’s biggest threat or its biggest growth engine, depending on the rules we set today,” Abbas said. She described the technology as “a guest we didn’t invite” that is “already in our house, sitting on the couch, flipping through our books.”
The AI market in publishing is projected to reach $3.7 billion by 2030, according to Abbas, who noted that artificial intelligence is “running through every part of the industry: generating content, translating across languages, enriching metadata, optimizing distribution, and even managing rights.” HarperCollins is seeking to balance intellectual property protection with the opportunities to monetize content—and viable models need to emerge. “It’s essential to change the narrative from we own the content to license the content legally,” Restivo-Alessi said.
The company has experimented with AI applications including chatbots as book companions, AI-enhanced translations, and tools to identify books to be republished from HarperCollins’ 200-year backlist. “AI is a huge equalizer,” Restivo-Alessi argued. “Smaller publishers are often faster and more adaptive because they don’t have 50 people to coordinate. This is an empowering moment for smaller publishers.”
Legal challenges are mounting against AI companies across multiple jurisdictions. In the United States, authors including Sarah Silverman have filed lawsuits against AI developers over copyrighted works in training datasets. European publishers are advocating for stronger protections under the EU’s AI Act, while collective lawsuits proceed in Germany and France.
Global regulatory approaches
Dr. Jessica Sänger outlined the varying global regulatory approaches. The European Union has implemented two exceptions for text and data mining, allowing scientific research organizations to use copyrighted works and permitting broader mining activities with an opt-out mechanism for rightsholders.
“If rightsholders can manage these rights well and license them effectively, it helps prove to policymakers there is a market here that they shouldn’t remove by introducing broad copyright exceptions,” Sänger said.
Industry concerns extend beyond direct copyright infringement to anti-competitive effects and market dilution. Publishers worry about AI companies developing competing products using their content without authorization, particularly in educational publishing and companion applications.
Concerns about direct competition
“The main concerns include mass infringement of copyright by ingest for AI training and other purposes,” Sänger said. “There’s also concern about direct competition—if somebody else uses your works first because they were faster or more brazen.”
Publishers have developed technical standards including TDM Rep, a machine-readable rights reservation protocol for text and data mining that allows publishers to signal availability for licensing discussions while blocking unauthorized scraping.
The industry response varies significantly. Some publishers are adding “no AI use” clauses to contracts, while others actively pursue licensing negotiations with technology companies. Most remain conflicted between protective instincts and revenue potential.
“If you’re not in the conversation, you can’t shape the conversation, which is much worse. The value of rights is only increasing, it’s not decreasing,” Restivo-Alessi said.
