Dawn breaks after a sleepless night filled with shelling. The first rays of sunlight pierce through smoke-covered skies as a new, fateful day begins, opening another page of existence for a person who survived the night's horror and stands ready to continue their journey along a new threshold of life. The body, much like the soul, feels incredible exhaustion, yet the mind finds no peace, seeking solace in the familiar aspects of its daily routine.
Thus begins the morning of an ordinary Ukrainian translator who, wishing to distract herself from the stress she has just endured, immerses herself in her work and opens her laptop. She types a prepared prompt into her usual neural network alongside a fragment of a poem whose lines stubbornly refuse to find a concise rhyme in her head. Naturally, the machine responds almost instantly. The result turns out imperfect - a bit dry, somewhat artificial, slightly naive, and devoid of the pure breath that animates true poetry. Yet within this generated material emerges the exact euphonic resonance that the translator would have spent endless days searching for on her own. The job is done, and with a clear conscience, she uses the AI-generated material, discards the rest of the output, and rewrites the line anew - this time guided by her own hand, her own pain, and her own lived experience.
A question that just a few years ago would have sounded like a piece of science fiction has now become our everyday reality, forcing us to ask: who is the true author of that generated line, and is there an author here at all in the sense that world culture has ascribed to the word for centuries?
To begin with, it is worth emphasizing that artificial intelligence has long ceased to be something exotic, not only in programming but in our daily lives as well. It is built into text editors for auto-suggestions, into widely used translation software, into graphic applications, and into various digital services that have long become part of both digital culture and public life as a whole.
According to some estimates released at the beginning of this year, nearly a third of all new Ukrainian music released in 2025 across various platforms was created with the help of AI tools, pointing to the deep integration of these technologies into our artistic community.
For Ukrainian culture, this is not an abstract technological narrative; it is an everyday tool used by journalists, writers, translators, illustrators, and composers - individuals whose creative work, like that of most Ukrainians, takes place under conditions that can hardly be called normal, amid constant shelling, power outages, and a state of perpetual internal mobilization. Yet the question arises: do these circumstances justify abandoning individual creative endeavor in favor of artificial generation?
The question has remained open for quite some time, as many creators do not perceive artificial intelligence as a threat from the future, but rather view it as an everyday assistant. It helps them cope with burnout, overcome the paralysis of the blank page, and speed up routine tasks to preserve energy for the work that only a human being can accomplish.
Yet in practice, this is precisely where the greatest danger lies - he blurring of the boundary between creation and generation, between human contribution and algorithmic output. This boundary carries concrete legal and financial consequences, naturally raising the question: who should receive royalty payouts for these works, whose name should appear on the finished piece, and who holds the right to authorize or prohibit its further use?
The answers to these questions stem not from ethics or intuition, but from a concrete and well-established field of law: intellectual property law.
Before discussing artificial intelligence, it is worth recalling the foundation upon which copyright rests - a framework that is over a century old and until recently seemed so self-evident that few ever dared to question it.
The Law of Ukraine "On Copyright and Related Rights" No. 2811-IX of December 1, 2022, in force since January 1, 2023, provides an exceptionally straightforward definition in Article 1, emphasizing that an author is a natural person who created a work through their creative activity. Thus, an author is neither a company, nor a software program, nor an algorithm, but specifically a natural person. In turn, a work is defined as an original intellectual creation of an author in the field of science, literature, or art, expressed in an objective form.
Two words carry the entire weight of this structural pillar: "original" and "creation."
Originality in the global legal tradition of copyright is not merely novelty or the absence of copying; it is a true reflection of the author's personality - the imprint of their unique, free choices made from a multitude of possible forms of expression.
A work is traditionally viewed as something into which a human leaves a piece of their own soul, rather than a mere informational object that came into being on its own.
This idea was not invented yesterday. The Berne Convention for the Protection of Literary and Artistic Works, which is over one hundred and thirty years old and whose inception was personally guided by the great writer Victor Hugo, nowhere explicitly defines the word "author." For the creators of text in 1886, doing so would have seemed like a tautology: an author, naturally, could only be a human being, and anything else was simply nonsense. International intellectual property law was built for centuries on this silent, unspoken presumption. Generative artificial intelligence became the first technology in history to force this presumption to speak aloud and prove itself.
In this regard, Ukraine is one of the few countries in the world that has already attempted to give generative artificial intelligence a direct statutory response, rather than leaving the matter entirely to judicial practice.
Article 33 of Law No. 2811-IX introduced into Ukrainian law a concept known in Latin as a *sui generis* right - a right of a special kind granted to non-original objects generated, as the legislature framed it, by a computer program. Under "computer program," the legislature here refers, among other things, to artificial intelligence systems such as large language models familiar to us all.
In the law itself, this definition is rather narrow, stipulating that a non-original object is one formed as a result of the functioning of a computer program without the direct participation of a natural person in its creation.
The definition is quite clear, but it is immediately followed by a crucial saving clause upon which the whole framework rests: works created by natural persons using computer technologies are not considered non-original objects. This demonstrates that the mere use of AI determines nothing on its own; what matters is the degree and nature of human participation.
What does this *sui generis* right grant to its holder? Truth be told, significantly less than full copyright protection.
First, it creates no moral rights, meaning it grants no authorial rights to attribution, to the integrity of the work, or to the protection of the creator's reputation. The *sui generis* right is purely economic and commercial in nature.
Second, its term of protection is far shorter than that of ordinary copyright, which in Ukraine - as in most states worldwide under international conventions - lasts for the author's lifetime plus seventy years after their death. In contrast, the validity of a *sui generis* right expires after twenty-five years, calculated from January 1 of the year following the year the object was generated.
Furthermore, the most intriguing question that frequently perplexes those reading Article 33 for the first time is: who exactly owns this right?
Intuition suggests it should belong to the person who submitted the prompt - the author of the request to the artificial intelligence. While that might seem obvious, the law tells us otherwise. The legislature designates as subjects of this special right the persons holding economic rights to the computer program itself or those possessing licensing authority over it. This includes the original developers of the program and their heirs, assignees of economic rights, or lawful users of the computer program. Crucially, the law explicitly states that the terms of ownership may be determined by contract, meaning everything hinges on the terms of service set by the AI developer in their user agreement. Someone who simply pressed "generate" and published the result without any changes may turn out not to own the results of their interaction with the machine at all, if the terms of the service they used stipulate otherwise.
If we distill all of the above into a single practical principle, it looks like this: the law - both in Ukraine and in almost every other jurisdiction worldwide - draws the boundary not between "human" and "artificial intelligence," but between artificial intelligence as a tool and artificial intelligence as an autonomous generator.
While this may sound somewhat counterintuitive and complex, let us examine this framework through a concrete example.
Imagine three creators. The first uses artificial intelligence as a sounding board to discuss the structure of a future literary work, but then sits down and writes every sentence himself, in his own language, with his own imagery and factual detail. The second inputs a prompt, receives a paragraph, and publishes it without a single edit. The third, an illustrator by trade, generates a dozen image variants in a graphic application, selects details, arranges them into a new artistic composition, and manually draws and retouches what was lacking in her vision.
In the first scenario, we are dealing with a full-fledged work in the classic copyright sense, complete with all economic and moral rights, protected for the full seventy years following the author's death. In the second scenario, the output is most likely a non-original object covered by Article 33 - something that can be used, but which is not a "work" in the legal sense and does not automatically belong to the person who published it. The third scenario is the most interesting and common in practice, as the outcome depends on the extent to which the author's selection, arrangement, and manual refinement outweighed the initial machine-generated material, transforming it into something bearing the imprint of a person's individual creative choice.
It is in this middle ground that the main legal debate is unfolding today, not only in Ukraine but across the globe. Turning to international experience reveals that despite differences in legal terminology, all major legal systems are grappling with essentially the same problem.
It is important to note that at the European Union level, there is no standalone law on copyright for AI-generated objects equivalent to Ukraine's *sui generis* right.
Instead, for over fifteen years, the EU has relied on a general test formulated not by the legislature, but by the Court of Justice of the European Union (CJEU). This test establishes that a work is eligible for protection only if it constitutes, in the Court's wording, the author's "own intellectual creation."
The CJEU established this standard in the landmark 2009 *Infopaq* case, subsequently developing it across a series of rulings.
In the 2011 *Painer* case, which concerned an ordinary portrait photograph, the Court emphasized that even when an author works with a technical tool - in that instance, a camera - originality remains possible if the person makes free and creative choices regarding angle, lighting, timing, and composition. Later rulings in *Levola Hengelo* (2018) and *Cofemel* (2019) definitively affirmed that only those elements of a work that genuinely reflect the author's personality and can be identified with sufficient precision are eligible for copyright protection.
It is easy to see how the logic of the *Painer* case applies to working with artificial intelligence. A camera is also a technology that generates an image according to physical laws of optics rather than creative intent; protection arises not from the mechanical press of a button, but from the creative decisions made before and after that press.
Accordingly, European legal scholars today interpret human interaction with generative models as working with a tool - regardless of how complex that tool may be. Legal protection arises where there is room for free human creative choice, echoing the philosophical paradigm that a human remains human so long as they are guided by their own freedom of choice.
The European Union's Artificial Intelligence Act (AI Act), passed in 2024, does not directly regulate copyright authorship, but it takes a major step toward transparency. It obliges developers of general-purpose generative models to make publicly available a sufficiently detailed summary of the copyrighted materials used to train their systems. This provides rightsholders - including Ukrainian writers and illustrators whose works may have been swept into training datasets - with at least a theoretical mechanism to verify whether their rights were infringed, allowing them to exercise the opt-out rights provided under the Digital Single Market Directive (2019/790) to prohibit the use of their works for machine learning.
While Europe relies on judicial interpretation of existing concepts, the United States has chosen a path of explicit, almost categorical declaration: without a human, there is no author!
The central story here is the multi-year legal saga *Thaler v. Perlmutter*. While the case name may mean little to the general public, its details are remarkably compelling.
It began when computer scientist Stephen Thaler created an AI system named the "Creativity Machine," which generated a visual piece titled *A Recent Entrance to Paradise*. Thaler submitted an application to register copyright for the image, listing the machine itself as the author and describing himself merely as the owner of the rights transferred from the machine. The U.S. Copyright Office denied registration, and Thaler challenged the refusal in court.
On March 18, 2026, the U.S. Court of Appeals for the District of Columbia Circuit definitively upheld the Copyright Office's position, ruling that human authorship is a "fundamental requirement" of copyright law and that U.S. copyright protects exclusively the fruits of human creativity. In May of the same year, the court declined to rehear the case *en banc*.
This ruling aligns with the conclusions of the U.S. Copyright Office itself, which in January 2025 published Part 2 of its comprehensive report on artificial intelligence and copyright.
The report's main conclusion was that inputting a prompt alone - no matter how detailed or imaginative - does not constitute sufficient human authorship to warrant protection for the generated output. At the same time, a process in which a human selects, edits, arranges, or creatively transforms the raw output can give rise to protected authorial work, at least regarding those elements where human creative choice was expressed.
A telling illustration of this principle is the internationally known case of the comic book *Zarya of the Dawn*. The author utilized images generated by the Midjourney neural network and sought to register the AI as a co-author. The Copyright Office initially granted registration, but later canceled it upon review. Copyright protection was retained for the text written by the author herself and for her creative selection and arrangement of the images into a narrative sequence, but protection was stripped from the individual generated illustrations on the grounds that they lacked sufficient human authorship.
This case serves as a striking example of the tool-versus-author distinction applied to a concrete, highly practical scenario.
There is, however, a fundamentally different approach demonstrating that global consensus is lacking even on basic principles.
Since 1988, UK copyright legislation has contained provisions governing "computer-generated works" - defined as works generated by a computer in circumstances where there is no human author in the traditional sense. In such cases, the law does not leave the work unprotected; instead, it explicitly designates as the author the person by whom the arrangements necessary for the creation of the work are undertaken.
This is a rare global example of a legislature legally recognizing the organizer of a creative process as a full-fledged author, albeit without moral rights and with a reduced term of protection. Passed long before the advent of modern generative models, this provision is now forcing British lawyers and lawmakers to re-examine whether it remains suitable for the era of large language models or requires revision.
Comparing these four national approaches highlights the distinct strategies currently in play. Ukraine chose the path of a special, time-limited *sui generis* right without moral rights. The United States chose categorical refusal of protection wherever human creative contribution is absent. The European Union adopted flexible judicial interpretation of traditional originality criteria, while the United Kingdom opted for assigning authorship to the entity arranging the creative process. As a result, we observe four distinct legal mechanisms driven by a shared intuition that each system attempts to safeguard in its own way: value must follow human decision-making, rather than the mere mechanical emergence of text, imagery, or music.
At the global level, no unified binding standard currently exists. The World Intellectual Property Organization (WIPO) has engaged in a structured dialogue on this topic for several years within the framework of the *WIPO Conversation on Intellectual Property and Frontier Technologies*.
The April 2025 session in Geneva focused directly on the role of the human author in the age of artificial intelligence, the October 2025 session addressed synthetic media, and in March 2026, WIPO launched a dedicated platform for technical dialogue between AI developers and rightsholders. In autumn 2026, WIPO will host its first Global Forum on Intellectual Property and Artificial Intelligence. However, these initiatives remain forums for discussion rather than venues for adopting binding international norms or the legally enforceable treaty desperately needed by the global community to regulate AI-assisted authorship - a treaty that does not yet exist and is unlikely to emerge in the immediate years ahead.
Returning to Ukrainian legislation, it is vital to emphasize that Article 33 of Law No. 2811-IX is a valuable yet narrow instrument. It addresses a specific copyright issue but does not constitute comprehensive regulation of artificial intelligence - a regulatory framework that Ukraine currently lacks entirely.
This situation is expected to change by the end of this year. The Ministry of Digital Transformation announced that throughout 2026 it will work on a draft law dedicated to artificial intelligence, intended to harmonize Ukrainian legislation with the European AI Act.
This step is logically necessary given Ukraine's status as a candidate for European Union membership, for which aligning national law with the *acquis communautaire* is a formal requirement rather than a mere suggestion.
By mid-July 2026, preparatory work was already complete. A specialized working group comprising representatives from business, academia, and education prepared a roadmap and a White Paper on AI regulation in Ukraine, selecting a strategy that encourages business self-regulation alongside gradual adaptation to future rules. In autumn 2025, a national AI assistant went live on the Diia portal, showcasing how the state itself is integrating these technologies into public administration while declaring an ambitious goal to rank among the top three global leaders in government AI implementation by 2030.
Yet even after new legislation is enacted, questions will remain that Article 33 cannot answer.
For instance, how do we define the lawful user of a computer program when dealing with a global platform registered abroad and subject to foreign law? How can we ensure that foreign developers of large language models respect Ukrainian law when training their systems on text datasets that may include works by Ukrainian authors? Landmark lawsuits are already unfolding worldwide - such as actions brought by Getty Images or groups of artists against generative model developers - seeking to resolve the fundamental question of whether training neural networks on protected works without rightsholder consent is lawful in the first place.
Ukrainian artists, authors, and publishers must closely monitor these legal developments, as they will inevitably impact them if they have not already.
All of this could be left to legal specialists if it were merely a matter of technical registration procedures. But for Ukrainian culture - currently operating in a state best described as survival - this debate is no abstract exercise; it is a matter of economic and moral survival for an entire professional community.
Throughout 2025, Ukrainian literary translators, alongside colleagues across the cultural sector, repeatedly engaged in public discussions regarding the threat artificial intelligence poses to their profession. They highlighted the subtle nuances of literary translation that machines cannot replicate: tone, cultural context, and the courage to make a bold interpretive choice where a literal translation would be dead on arrival. Numerous Ukrainian and international figures have noted that while AI can serve as a helpful tool for authors, students, and researchers, it cannot replace a writer because it lacks lived human experience. It continues to make errors, particularly on niche topics, occasionally hallucinating non-existent facts. At the same time, distinguishing machine-generated text from human writing based on surface appearances alone is becoming increasingly difficult.
The international *ARTIFACT* project, presented with the participation of Ukrainian artists, offered a poignant look at the intersection of art and AI through the lens of wartime culture. Ukrainian participants underscored the necessity of establishing shared transparency rules: if an artist uses AI tools in their work, they should clearly indicate which portion was performed by the algorithm and which was executed by the artist.
They also emphasized that while a machine can serve as an instrument or assistant, it cannot replace the human soul behind a work of art.
There is also a direct economic dimension to this issue that intersects with the work of Collective Management Organizations (CMOs). These organizations, already navigating the strain of war, accreditation reforms, and legislative shifts, face yet another practical hurdle: how to distinguish a human-composed piece from a neural-network-generated melody when both enter the same catalog for royalty distribution. If the law and legal practice fail to draw this line clearly and consistently, the market risks being inundated with cheap, characterless, non-original content. This flood threatens to displace the paid labor of living poets, translators, illustrators, and composers - many of whom are already creating under exhausting and dangerous conditions, yet continue to work, build, and produce our national heritage. Drawing this legal line with precision and care is a matter of basic economic survival for a fragile industry.
Several practical takeaways emerge from this analysis for anyone who writes, translates, draws, or composes music daily.
First and foremost, every modern creator must familiarize themselves with relevant legal standards within their field. Read the terms of service of the AI platforms you use. Article 33 of Law No. 2811-IX explicitly allows contracts to determine who holds the *sui generis* right to a generated object, meaning a platform's user agreement may claim broader rights than an average user might intuitively expect.
Additionally, document your creative process. Save drafts, keep logs of your prompts, and, most importantly, retain records of your own edits and manual refinements made over machine-generated material. This paper trail of creative choice - rather than the mere fact of whether AI was used - will serve as decisive evidence of authorship should you ever need to defend it.
For publishers, contracts with authors must clearly differentiate between two fundamentally distinct scenarios: a human work created using AI as an auxiliary tool, and a non-original object generated entirely by a machine.
Crucially, creators should follow the legislative progress of the draft AI law being prepared by the Ministry of Digital Transformation, as it may alter the rules of the game. Professional and creative unions ought to actively advocate now to ensure that the value of human creative labor receives unambiguous protection under the new statute.
Ultimately, let us return to the translator from the opening line of this article. She drew inspiration from the machine and set everything else aside. It was in that active gesture of selection - in finding the precise cadence that resonated with her own artistic ear - that the essence of what intellectual property law strives to describe and protect truly resides.
The word "muse" in European culture has never referred to a co-author. A muse does not write the poem for the poet; she merely inspires him to reach beyond what he thought himself capable of achieving. In its healthiest and most honest application, artificial intelligence occupies precisely this role in the creative process: the role of a muse, not an author.
Ukraine's *sui generis* right for non-original objects, the American fundamental requirement of human authorship, the European standard of intellectual creation, and the British framework for the organizer of the creative process - all these varied legal concepts speak to the same fundamental truth: rights and royalties must follow human decision-making, human experience, and human choice, rather than the mere mechanical generation of text, imagery, or melody.
This boundary is delicate, and maintaining it will grow increasingly challenging as algorithms become ever more convincing at mimicking the human voice, style, and thought process. Yet that is precisely why each of us must draw this line consciously, carefully, and with a clear understanding of what is at stake. What lies on the line is not abstract legal elegance, but the real lives of ordinary people - like our translator searching for the right word on a morning after an attack, and thousands like her for whom Ukrainian culture today represents not merely a profession, but a vital form of spiritual and intellectual resistance. A digital muse may suggest a line, but at the end of the day, the name beneath the poem must belong to the human being who dared to speak something genuine, living, and true.
Author: Hryhorii Khoroshko
