Artificial intelligence could create a new copyright and licensing area in publishing

Claims that artificial intelligence companies buy printed books in bulk, scan them, transfer them to digital and then destroy them, attracted the reaction of second-hand booksellers in Europe, while similar situations began to occur in different countries. While bulk orders, especially for academic publications and rare works, raise suspicion, the unethical attitudes of companies ordering books from Türkiye are discussed.

Halil Çelik, President of the Press and Publication Association, evaluated the latest developments in the publishing sector in the last news file of Anadolu Agency (AA).

Çelik said that we will face an important process regarding the content used in the training of artificial intelligence models in the near future.

Stating that the creation of a book requires economic investment with the efforts of writers, translators, editors, publishing houses, designers and other stakeholders, Çelik said, "Using these contents in the training of artificial intelligence systems without any permission and fee mechanism actually means using the economic value of this labor for free. The main problem here is not just the copying of the book. The contents that the publisher has created over the years and has commercial value become the input of the commercial models developed by artificial intelligence companies. Moreover, today there is an adequate mechanism to allow the publisher to know which book is used in the training of which model and to what extent." "It does not exist." he said.

Stating that the possibility of digitizing old and rare books to be used in the data sets of artificial intelligence companies is also worrying, Çelik said that some works can be physically disassembled to be scanned.

"It is alarming for rare and out-of-print works."

Halil Çelik pointed out that cultural heritage is in question, especially in terms of works that are no longer published, and said:

"If these books are transferred to the digital environment to be used in the data sets of artificial intelligence companies, the books may need to be physically torn apart, their pages scanned one by one and digitized in order to be scanned. This is especially worrying for rare and no longer printed works. Because perhaps one of the last few copies in a library or on the shelf of a second-hand bookstore can be physically destroyed by being destroyed during the digitization process. Thus, while the content is digitally reproduced, on the other hand, a physical copy that is a cultural heritage can disappear."

Explaining that the main expectation from publishers is transparency regarding the use of content and a fair licensing mechanism, Çelik said that artificial intelligence education, especially based on entire books or large data sets, may turn into a new licensing area for publishers in the future.

"Licensing and royalty payment model should be established"

Stating that the reproduction, processing, dissemination and other financial rights of the author are protected in the current legal framework, Çelik emphasized that there is no special regulation regarding the use of works as a data set in the training of artificial intelligence models.

Stating that this situation creates an important legal gap and field of interpretation, Çelik said, "Artificial intelligence companies should explain the contents they provide training with to a certain standard. A licensing and royalty payment model should be established for commercial artificial intelligence training. This does not have to be in the form of an individual contract for each work. A system that can provide collective licensing on behalf of publishers, authors and other rights holders can also be developed." made his assessment.

Underlining that use for artificial intelligence education should be clearly defined in the Law on Intellectual and Artistic Works, Çelik said that the right holders' ability to grant permission or determine the conditions of use should be guaranteed, an obligation of transparency should be imposed on the source of educational data, and an effective sanction mechanism should be established in case of violation.

Halil Çelik stated that such a regulation will both protect rights holders and clarify which rules artificial intelligence companies will act according to, and continued as follows:

"Our aim here is not to prevent the development of artificial intelligence, on the contrary, we want the publishing industry to benefit from the opportunities offered by artificial intelligence. However, while this technological transformation takes place, the labor and economic value of the author, publisher and other rights holders must be protected. We believe that as artificial intelligence develops, the publishing sector should not be left out of this development and the value arising from the content it produces should be shared fairly."

"A reliable data infrastructure is needed"

Stating that a new licensing economy should be formed in the future rather than a conflict between artificial intelligence companies and publishing houses and copyright owners, Çelik said that artificial intelligence needs quality content and the publishing industry needs the protection of the economic value arising from the production of these contents.

Emphasizing that the industry should be prepared for this process, Çelik pointed out the importance of regularly recording the contents, rights holders and terms of use of publishing houses.

Halil Çelik emphasized the importance of acting together as a sector and said, "The negotiation power of individual publishing houses may be limited against large technology companies. Publisher associations, author organizations and other rights holders should develop a common model. There is a need for a reliable data infrastructure on which work belongs to whom, who has which rights, and whether it can be licensed for artificial intelligence education." he said.

Stating that issues such as scope of use, duration, region and pricing should be clearly determined in licenses to be made with artificial intelligence companies, Çelik noted that a transparency and control mechanism regarding the use of content should also be established.

Çelik concluded his words by underlining that artificial intelligence should not be seen as a threat in terms of publishing: