Ethical Generative Art in the Age of Creative Algorithms
AI generative art raises crucial ethical questions: copyright, originality, artist rights. The debate between human and algorithmic creativity.
Open MidJourney, type "impressionist sunset over Venice in Monet's style," wait thirty seconds, and get a breathtaking image that looks painted by a master. Upload it to Instagram, receive hundreds of likes. But is that really art? And if so, whose is it? Yours, the algorithm's, or the artists whose work trained the system?
These questions are not academic. They are at the heart of legal battles, ethical debates, and the very redefinition of what it means to be creative. Ethical generative art is the challenge of our time: drawing boundaries in a world where algorithms create works that move, provoke, and, increasingly, sell.
What is Generative Art and Why Does It Need an Ethic
Generative art is that created by autonomous systems, typically artificial intelligence algorithms, which produce visual, musical, literary, or multimedia works based on instructions, data, or prompts provided by a human. It's not entirely new: as early as the 1960s, artists like Harold Cohen with his AARON program explored computational creativity. But what we call generative AI art today is something profoundly different.
Models like DALL-E, Stable Diffusion, MidJourney, or Runway ML have been trained on billions of images downloaded from the internet, often without asking permission from the original artists. They learn styles, techniques, compositions. Then, when you enter a prompt, the algorithm does not "copy" a specific work, but synthesizes learned patterns to generate something new. Or at least, that's what the makers of these tools claim.
The ethical problem arises right here. As explained in the academic study published on arXiv, the implications of generative AI in creative industries touch on intellectual property, environmental impact (training these models requires enormous amounts of energy), deepfake risk, the disappearance of creative jobs, and the urgent need for updated regulation.
Art has always had an ethical dimension: who can create? Who owns the work? How is value recognized? But with AI, these questions multiply because the line between tool and creator blurs. A brush does not decide what to paint. An algorithm, in a sense, does.
To better understand this evolution, it's worth reading our article on how AI is transforming art and creativity, which explores the foundations of this cultural revolution.
Can an AI-Generated Work Have Copyright?
Here we are at the beating heart of the debate: authorship. In traditional intellectual property law, copyright protects works of human intellect. This seemingly simple definition becomes a minefield when AI enters the picture.
In the United States, the U.S. Copyright Office recently concluded that artists can obtain copyright for works created using AI as an "assistive tool," but not for works generated autonomously by the algorithm without significant human creative intervention. It's a subtle but crucial distinction.
What does "assistive tool" mean? If you use Photoshop with AI filters to retouch a photo you took, you are the author. If you use AI to generate variants of a sketch you drew, you probably still are. But if you simply write a prompt and the algorithm does everything else, then who is the author? The current legal answer is: no one. The work falls into the public domain.
This was confirmed by a ruling from a U.S. Appeals Court that denied copyright for AI works without substantial human involvement. The case involved Stephen Thaler, who had attempted to register copyright for a work generated entirely by an AI system called "Creativity Machine." The court ruled that "human authorship is a fundamental requirement for copyright protection."
Our article AI and Copyright: Who Owns the Work? delves into these legal dynamics with concrete examples and analysis of different jurisdictions. Furthermore, we have documented three real cases of copyright infringement with AI that show just how complicated the current situation is.
The Artists' Perspective: Fear, Anger, and Demands
While legal experts and legislators try to figure out what to do, flesh-and-blood artists are living this revolution firsthand. And they are not happy.
An extensive survey published on arXiv, involving over 400 professional artists, reveals alarming data: 73% of respondents fear that generative art will drastically reduce job opportunities. 68% believe their styles have been "stolen" without consent to train commercial models. Only 12% see AI as a positive opportunity for their career.
Artists' concerns focus on three key points: transparency (they don't know if or how their works are being used), ownership (they receive no compensation for the use of their work in training), and fairness (they compete with systems that reproduce their style in seconds, at zero cost).
Take the case of Greg Rutkowski, a Polish digital illustrator specializing in fantasy art. In 2022, he discovered his name was among the most used prompts in Stable Diffusion. Thousands of people were writing "in the style of Greg Rutkowski" to get images in his characteristic style. Without ever asking his permission. Without giving him credit. Without paying him a cent.
Or consider the class action lawsuit filed by Sarah Andersen, Kelly McKernan, and Karla Ortiz against Stability AI, Midjourney, and DeviantArt. The artists allege that billions of their works were used without consent to train the models, constituting a massive copyright violation. The lawsuit is still ongoing and could redefine the entire industry.
As Data Science Central reports, the boundary between human and automated creativity is becoming increasingly blurred, raising not only legal but deeply ethical questions about ownership, biases embedded in datasets, and the responsibility of those developing these systems.
The theme of human versus algorithmic creativity is so central that we also discussed it in the article AI Artist: Friend or Foe of Creativity?, where we explore the different points of view.
Originality and Authenticity: The Philosophical Debate
Beyond the law, there is philosophy. Can a work of art generated by an algorithm truly be original? And what does "authenticity" mean when the creator has no emotions, experiences, or intentions?
Human art stems from lived experience. An artist paints a sunset because they have seen it, felt it, experienced something. The algorithm feels nothing. It synthesizes statistical patterns. When MidJourney generates an "emotionally intense" image, it is not feeling emotions: it is replicating visual combinations that humans have historically associated with certain emotions.
As analyzed by Visual Alchemist, the debate on the originality of generative art intertwines with age-old questions: what is creativity? Is it the ability to do something new, or is it inseparable from consciousness and intentionality? If an algorithm produces an image no human had ever imagined, is that original?
Some argue that originality lies not in the object but in the process. Human art is original because it derives from a unique and unrepeatable subject. Algorithmic art, even when it produces new results, stems from a replicable and deterministic (or pseudo-random) process.
Others counter that human artists also "copy" and synthesize. Picasso said: "Good artists copy, great artists steal." Every artist is influenced by those who came before them. In this sense, AI does nothing qualitatively different: it learns from examples and produces variations.
But there is a difference: the human artist consciously chooses what to imitate, what to reject, what to transform. The algorithm does not. It processes everything indiscriminately, without understanding, without taste, without ethics.
This reflection connects to our article on augmented literature and AI as a co-author, where we explore similar dynamics in the field of creative writing.
Ethical and Unethical Use Cases of Generative Art
Not all generative art is equal from an ethical standpoint. Let's make some practical distinctions.
Ethical Use:
- A graphic designer uses DALL-E to generate initial concept drafts, then manually reworks them
- A disabled artist uses AI to compensate for physical limitations, while maintaining creative control
- A designer uses AI to explore color or composition variants of their original work
- An artist openly declares the use of AI and credits the authors of the datasets when possible
Problematic Use:
- A company replaces human illustrators with AI-generated images to cut costs
- Someone sells prints generated with prompts "in the style of [living artist]" without consent
- A social media influencer presents AI artworks as their own original creations without disclosure
- A brand uses AI art trained on copyrighted works for commercial campaigns without licenses
As highlighted by ArtsHub Australia, the key lies in transparency, acknowledging social impacts, and managing the biases embedded in training datasets.
A positive example comes from Refik Anadol, a Turkish-American artist specializing in data art. Anadol uses AI to create immersive installations based on large datasets (museum archives, environmental data), but is always transparent about the process, collaborates with cultural institutions, and conceives of AI as a creative collaborator, not a replacement.
A negative example? The case of AI-generated book covers on Amazon KDP. Thousands of self-published authors have started using MidJourney to create covers, often imitating the style of professional illustrators. Result: the market for commissioned covers has collapsed, while AI works saturate the platform with variable aesthetic quality.
To explore the creative implications in other fields as well, I recommend our article on how AI is transforming music, where similar dynamics are emerging.
Towards ethical guidelines for generative art
If the Wild West of AI art is unsustainable, what rules are needed? Various organizations and institutions are working on ethical frameworks.
Emerging principles:
Mandatory transparency: Anyone using AI to create art should declare it explicitly. Some propose a permanent digital "watermark" for AI-generated works, as analyzed by the Michigan Tech University blog.
Opt-in for artists: Training datasets should only include works whose creators have given explicit consent. This flips the current approach, which is opt-out (the artist must explicitly request exclusion) or, worse, offers no option.
Fair compensation: If a model earns billions using the work of millions of artists, they should receive a share of the profits. Models similar to those of music rights management societies (SIAE, ASCAP) could be adapted.
Style Control: Living artists should be able to prohibit the use of their name or recognizable style in commercial prompts. It's like the right to one's image, but for artistic style.
Certification of Human Intervention: To obtain copyright, one should demonstrate significant creative human contribution, not just the prompt. This would incentivize the use of AI as a tool, not a replacement.
As HAI Lu proposes, shared guidelines are needed among stakeholders (artists, tech companies, legislators) for responsible use that balances innovation and the protection of rights.
The Modern Diplomacy article offers a critical view on how generative AI is influencing authenticity, cultural values, and the art market, urging a deeper reflection on the long-term consequences.
This debate is part of a broader theme we explored in the article on the AI Moral Code, where we examine the fundamental values that should guide the development of artificial intelligence.
📌 Key Points to Remember
Copyright Requires Human Creativity: Currently, in the USA and many jurisdictions, works generated entirely by algorithms without creative human input cannot be protected by copyright. AI can be a tool, but authorship remains human.
Artists Demand Protections: The majority of professional artists see generative art as a threat, not an opportunity. They demand transparency on how their works are used, fair compensation, and the right to veto the use of their style.
Ethics Goes Beyond Legality: Even if something is technically legal (like using public domain works for training), it can raise ethical questions. Cultural sustainability requires respect for human creative work.
Transparency is Fundamental: Disclosing the use of AI does not diminish the value of a work. Hiding it, however, fuels distrust and harms the entire creative ecosystem. Disclosure should be the norm, not the exception.
❓ FAQ
Can I sell works created with MidJourney or DALL-E?
It depends on the platform's terms of service and local copyright laws. MidJourney allows commercial use with some plans but does not guarantee that you have copyright over the work. In many jurisdictions, purely AI-generated works are not protectable. Always check the ToS and consider adding significant human creative modifications.
Is it ethical to use prompts "in the style of [living artist]"?
It is legally controversial and ethically questionable. Many artists consider it theft of creative identity. If you want to be ethical, avoid using the names of living artists without permission. Instead, describe the style with generic terms (e.g., "impressionist" instead of "Monet style") or use references to historical artists no longer under copyright.
How can I tell if a work was created with AI?
It's not always easy. Some clues: unnatural technical perfection, strange anatomy (hands with too many fingers, asymmetries), incomprehensible text in the image, an artistic signature that looks like gibberish. Some tools like Hive Moderation or Optic AI promise to detect AI art, but they are not infallible. The best solution remains to ask the author.
Will AI art replace human artists?
It will replace some jobs (stock illustrations, basic concept art, decorative images), but it is unlikely to completely replace human creativity. Art is not just the final product, but the process, the intention, the cultural context. AI can produce beautiful images, but it cannot (yet?) produce art with genuine meaning, original vision, and authentic emotional connection.
What can artists do to protect themselves?
Various strategies: use tools like Glaze or Nightshade that "poison" images, making them unusable for AI training; include licenses in the work's metadata that explicitly prohibit AI use; join associations that collectively negotiate with AI companies; support legislation requiring explicit opt-in; develop skills in the creative use of AI to not be replaced but enhanced.
The Art of the Future Will Be Hybrid
Ethical generative art is not an oxymoron. It is a design challenge. We can build an ecosystem where AI and human creativity coexist without one devastating the other. But it requires conscious choices.
AI developers must design systems that respect rights. Platforms must implement transparency and traceability. Legislators must update laws conceived for a pre-digital world. And we users? We must learn to distinguish, to ask, to value human work.
There is room for a future where the artist uses AI as an incredibly powerful assistant: the algorithm generates variants, the human chooses, modifies, gives meaning. AI accelerates the process, the human maintains the vision. This collaboration, if built on ethical foundations, could enormously expand creative possibilities.
The risk, however, is a world where art becomes an industrial commodity: mass production, zero cost, maximum efficiency, no soul. Where "artist" is no longer the one who creates, but the one who writes the best prompts. Where the museums of the future exhibit works generated in milliseconds, while human creators become archaeological relics.
Artistic deepfakes also raise similar questions about the manipulation of reality and authenticity in the digital age, showing how generative technologies impact different creative fields.
It is not inevitable. It is a choice. And that choice is being made now, in the courtrooms where copyright lawsuits are decided, in the parliaments where laws are written, in the tech companies where algorithms are designed, but also in our homes, every time we generate an image, buy a work of art, or support an artist.
Art has always reflected the values of the society that produces it. Which generative art we choose says a lot about which values we want to defend: efficiency or authenticity, speed or depth, consumption or connection.
The frontier is not only technological. It is human.