Author: Bilean Daniel Beraki
Publication date: 17.07.2024
Artificial Intelligence (AI) has rapidly transformed the way we create and consume content. AI-generated content now includes a wide range of media, from texts and music to art and videos. This technology has made content creation more accessible and efficient, allowing for new and innovative ways to produce and share creative works. However, the widespread use of AI in generating content has also raised important questions about copyright and intellectual property, as it becomes increasingly difficult to determine the ownership and originality of AI-created work.
What is Copyright?
Copyright law protects original works of authorship. Originality in this context means that the works owes its origin to the author, the work has to be independently created and not copied from other works. For a work to be copyrighted, it needs the minimal degree of creativity. Hoverer, this work needs to be an expression of an idea, it cannot just be ideas, concepts, principles or even facts and discovery.
A brief history of Copyright law
The evolution of copyright law dates back to significant milestones such as the Statute of Anne in 1710 and the Berne Convention of 1886. The statute of Anne marked the beginning of modern copyright law by recognizing authors’ rights to their literary works. This statute shifted the control of copyrights from publishers to authors as the primary beneficiaries. It also introducing the idea of the public domain, by utilizing the concept of limited-term protection once the copyright term expired, works could be freely used by the public.
Building on this, the Berne Convention established the minimum standards for the international protection of copyright. The convention introduced automatic protection, meaning that author’s right are recognized universally as soon as work is created. As well as national treatment and improved the minimum standards of protection for a wide variety of literally and artistic works across member countries. This developments have harmonized copyright laws globally, providing the most basic rights to creators across borders.
How is Copyright protected in the EU
The EU copyright law consists of 13 directives and 2 regulations, harmonizing the essential rights of authors, performers and producers. Many of these directives reflect Member States’ (MS) obligation under the Berne Convention. However, none of these directives offer a clear explanation of what constitutes an author’s work. Article 1 of the Term Directive tries to provide a general definition by describing copyright material as ‘a literary or artistic work’, as mentioned in Article 2 of the Berne convention.
Additionally, although the Court of Justice of the European Union (CJEU) references Article 2(1) of the Berne Convention for guidance, it bases its decisions on originality and identifiable expression rather than explicitly applying the domain test.
The EU standardizes three or four types of copy-right protected content: computer programs, databases, photographs, and possibly visual art. Each is protected if it is ‘the author’s own intellectual creation’. This was established as a general concept under EU law by the CJEU, in the 2009 Infopaq judgment, and was confirmed in later cases including Levola Hengelo and Brompton Bicycle.
While the EU law does not explicitly state that copyright requires a human creator, its focus on human authorship is evident through various aspects of the law. The CJEU case law on originality, relies entirely on the concept of human creativity involving ‘creative choices’. In the Cofemel case, the court reaffirmed that for a work to be deemed original, it must reflect the ‘author’s personality, manifested through their free and creative decisions’.
Finally, for work to qualify for copyright protection, the human creator’s creativity must be expressed in their final work. Reaffirmed by the CJEU in Infopaq and BSA, the creator must have ‘expressed his creative ability in an original manner’. This concept necessitates human involvement and a level of intentional creative effort.
Can AI work be protected under EU law?
Building on these requirements, AI-generated content lacking any human intellectual effort is not legible for copyright protection. However, can works created by human authors using machines be protected under copyright, provided that the human contribution to the output meets the legal standard of creativity?
AI generated content must meet the four step requirement to qualify as protected ‘works’: Firstly, it must be under the scientific, literary or artistic domain. Secondly, human intellectual effort must be involved in the creation. Thirdly, there must be a demonstration of originality in the creative choices. Lastly, the expression of the content must be detectable in the final product.
AI produces a variety of content including poems, music, art, pictures and more, satisfying the domain of ‘artistic, literary or scientific’ work. Additionally, at some level, AI-generated creation has some level of human input whether in developing the AI software, selecting training data, or refining the final product. The question then posed is even with the minimal human interaction, can AI generated work be protected by copyright.
The creative process in AI-assisted production is divided into: conception (design and specification), execution (producing drafts), and redaction (editing and finalizing). While AI dominates the execution phase, human involvement in the conception and redaction phase is crucial for ensuring sufficient creative input. If these human choices are expressed in the final output, it qualifies as work. Conversely, outputs generated entirely by AI without human creative intervention do not qualify.
Challenges and Legal Gaps
Currently, the only AI generated content eligible for copyright protection is that which includes human creative input. For example, if a person types in a question into ChatGPT, the output from that request can’t be copyrighted. However, if the author then makes their own creative change to it, that modified part can be protected.
The next sequential question is: What AI-generated content could be considered copyright infringement? AI generates content through computers programs that are trained on large amounts of existing texts. These programs, known as machine learning algorithms, learn to recognize patterns and common phrases in the data. Once trained, they can create new text that follows the same style or tone as the material they learned from.
The issue with AI-generated content is that users can’t verify if the AI is creating original work or copying form existing copyrighted material or other AI-generated content, leading to potential copyright concerns.
These advancements make it difficult to distinguish the true origin of the work. Who owns the intellectual property rights to outputs generated primarily by an AI system that was trained on thousands of existing books and journals, then reworded by a human author?
Traditionally, copyright laws grant ownership to the human, but with the significant help from AI ownership becomes unclear. Who really owns the content? The developers of the AI? The person who reworded it? Or potentially no-one if the work is deemed none-human generated. Ethical considerations and major legislative reforms are needed to clarify what is meant by an author and originality, as well as guidelines for disclosure of AI-assistance.
You can also read about:
תגובות