ETD

• AI Act
• artificial intelligence
• copyright
• digital market
• DLT
• European Union
• fair use
• GenAI
• governance
• levy
• licensing
• opt-out
• regulation
• remuneration
• technology
• training
• United States

The rapid development of Generative Artificial Intelligence (GenAI) has gradually intensified the debate around the lawful use of copyrighted works during the training phase of AI models. GenAI works by processing a large amount of data, usually obtained from the internet, and for this reason a vast amount of copyrighted content may be (even inadvertently) incorporated into training datasets. To properly regulate such a process, a fair and trustworthy remuneration system is required.
This thesis analyses the current proposals of fair remuneration mechanisms supported by Distributed Ledger Technology (DLT) to display which legal frameworks appear more adequate to protect authors’ rights and to guarantee a fair remuneration.
This work first establishes the theoretical foundations of copyright regulation, analysing the purpose of the authors’ remuneration and how it changes in the context of AI training, evaluating the solution of the opt-out mechanism and the role of collective management organizations and regulatory authorities.
It then provides a comparative analysis of the evolution of the legal framework governing copyright in the European Union and the United States, with particular attention to the Digital Single Market Directive, the AI Act, the U.S. fair use doctrine and recent legislative initiatives. The analysis identifies the legal gaps related to the GenAI context, evaluating the uncertain application of text and data mining (TDM) exceptions, the bonding between copyright protection and the physical territory, and the absence of harmonised rules able to guarantee equal treatment to the authors.
Building upon these findings, the thesis evaluates alternative remuneration models, from a legal point of view, including private levies, compulsory licensing and statutory licensing, considering whether a statutory licensing regime could actually represent the most balanced solution for reconciling innovation with the protection of authors’ economic interests.
Finally, some technical proposals of a DLT-based governance framework are analysed to assess whether and how they could be able to fit within the legal borders and their ability to enable transparency, auditable remuneration, and automation of the royalty distribution through smart contracts.
The results of the research will discuss key findings, highlighting both the potential advantages and limitations of DLT technology in this area, and conclude by summarizing the research and offering a forward-looking perspective on how this technology might be properly applied in copyright remuneration.