The use of books and other publishing materials by artificial intelligence companies for training commercial AI models without compensating copyright holders has sparked significant concern within Türkiye's publishing sector. Halil Celik, president of Türkiye's Press and Publishing Association (Basin Yayin Birligi), highlighted to Anadolu Agency (AA) that publishers currently lack a reliable means of tracking the books used in AI training and the extent of their utilization.
Read More
The controversy has escalated amid allegations that AI firms are purchasing physical books en masse, scanning them for digital datasets, and then destroying the hard copies. This practice has drawn criticism from secondhand booksellers across Europe, and similar issues have arisen regarding substantial orders of academic texts and rare books in various nations, including Türkiye.
Celik emphasized the particular risk to cultural heritage posed by the destruction of rare and out-of-print books, especially when companies scan pages from what may be the last available copies. He noted that while digitization can preserve content, it may result in the permanent loss of the original editions.
Beyond the preservation concerns, Celik pointed out that the publishing industry is also troubled by the economic implications of using their work without permission. When AI developers exploit these works unlicensed, the economic value generated by authors, translators, editors, and publishers is essentially used as input for AI systems.
To address these issues, Celik proposed that Türkiye's existing copyright laws be amended to better regulate the use of copyrighted works in AI training. He called for clearer guidelines regarding AI dataset disclosure, along with effective penalties for misuse. Additionally, he suggested the establishment of a licensing system that would require AI firms to pay for the commercial use of copyrighted material, potentially through collective agreements that encompass rights holders.
Celik believes that, despite these challenges, AI should not only be viewed as a threat. He argues that a well-regulated licensing market could potentially bring benefits to both technology developers and content creators. However, he acknowledged that smaller publishing houses often lack the negotiating power against large tech firms, which underscores the need for collaboration among publishers, authors, and other rightsholders.
He also advocated for a comprehensive database to identify copyright ownership and licensing eligibility, as well as tools for rights holders to monitor content usage. "If the right legal and licensing system is established, content that is currently exploited without payment could become a valuable source of revenue for rights holders in the future," Celik said, emphasizing the need to ensure fair treatment for all stakeholders in the evolving landscape of AI and published content.