Updated: 1 hour ago
Authorship and Ethics in the Collaborative "Grey Area"
Rachel Matthews
As an artist and art historian, I have always been fascinated by the intersections at which 'traditional' artistic heritage met emerging mediums or collided with societal change. Each new tool or style advancing (or igniting the downfall of) the craft.
The photograph. The cameras lucida and obscura. Closable paint tubes.
If we measure an artist by their ability to creatively conceive and technically produce novel works of art with the same tools we have traditionally considered appropriate or technically correct, it seems we're meandering down a dead end path. I would argue that an artist's challenge, simply, is to create something that conveys a message. The choice of medium is pivotal in both effect and the piece's ability to deliver that message. However, unlike the relatively recent emergence of Generative AI as a medium, its dismissal as a legitimate artistic tool and labeling it as requiring "no artistic skill" is unabashedly not new to the art world.
The Traditional European Art Establishment taught that true artistic genius required 'intellectual intervention'. To merely "copy nature" was considered vulgar, technically lazy, and lacking the intellectual ability required of a true artist. It was considered an artist's creative duty to seek beauty and to elevate, perfect, and moralize it through technical, stylistic, symbolic, and compositional methods. For hundreds of years even subject matter and compositions held strict hierarchical rankings. Rules were codified with the intention of exemplifying the artist's intellectually creative and technically accurate ability to create the most idealized works by current standards.
It's no surprise, then, that "not real art" isn't a new label. Johannes Vermeer was considered by many to be "cheating" when using a camera obscura to project shapes on his canvas to assist with capturing accurate perspective. Caravaggio was called lazy and said to have no talent for design because he often skipped preparatory sketches, opting instead to paint straight onto canvas from live models, sometimes with grids etched directly into the floor (lazy cheater!). Adolf Hitler was rejected from Vienna's Academy of Fine Arts twice as his works were too lifelike and lacking artistic imagination. Charles Baudelaire likened photography to a secretary (useful only for recording objective fact), assigning to it the death of the public's imagination for beauty. John Goffe Rand invented a collapsible tin paint tube in 1841 that allowed the for transportation and storage of oil paints that lead to their being used in new ways that were considered by many to be lazy and destroying of centuries of taught craftsmanship. Gutenburg's printing press was considered so heinous an invention that in pleading for it's ban, one friar warned that the invention "corrupts susceptible hearts" and "makes us excessively amorous". He made clear the dangers of countless books available for anyone to read; "the printing press is a whore" he wrote.
It's true that with change comes skepticism. Renoir credited Rand's paint tin with spurring the Impressionist movement. The printing press changed the way knowledge could be controlled and accessed around the world almost immediately, with an estimated 30,000+ new books published with as many as 20 to 25 million copies by the end of the century (guess it did get around).
To be clear- I'm not arguing that Generative AI is here to usher out the old way of the arts. Don't go tossing out your paints and pallets just yet.
It has, however, raised some age-old questions about artistic ability and responsibility, as well as intriguing new ones surrounding authorship, originality, and the role of the artist in a world where algorithms are capable of generating complex, compelling visual content.
We are living through one of the most significant societal shifts since the industrial revolution. Culturally, the rise of generative AI has sparked intense excitement and equally intense fear. Often, the conversation is reduced to extremes: AI is either the ultimate democratization of human creativity, or it is a weapon of cultural extraction. The printing press is a virgin or a whore. The truth, as always, is far more complicated.
Take the portrait above, for example.
The image is a product of my ideas and it exists because of actions I took, but it's far from my own original piece.
It draws deep stylistic inspiration from the palettes and philosophies of Frida Kahlo and Gustav Klimt. But rather than rendering this homage with traditional oils or even standard digital brushes I used Generative AI and several of my own photographs to create, adjust, tweak, and direct the composition, elements, colors, and iconography. Over and over.
The result sits squarely in the modern "grey area" of digital creation—a space that is currently fracturing the art world, testing the boundaries of copyright law, and forcing us to ask: What constitutes an artist’s tool, and what constitutes an artist?
If the artistic message, the conceptual heavy lifting, and the manual refinement are all driven by a human, does the fact that a machine rendered the underlying pixels diminish the cultural value of the work?
To many, the use of Generative AI to "create art" is culturally reductive. They cite that because these models are trained on billions of copyrighted works without consent, even the most granular, highly controlled prompting is inherently extractive—relying on the uncredited labor of the very historical masters I sought to emulate.
Others easily find this type of 'collaborative creation' to be a democratization of artistic execution. They argue that, much like the camera, AI removes technical rendering barriers, shifting the value of art from the physical discipline of how something is made to the conceptual weight of what is being communicated. A simple tool a creator uses to create.
As an artist I find the collaborative potential of this "grey area" genuinely fascinating. But as a professional who has spent close to two decades navigating the intersection of technology and photography, the infrastructural reality of how these models are built is glaringly exploitative.
What if we allow "scrape first, settle later" to become the industry standard?
When models are built on massive datasets of unlicensed work as corporations make the decision to simply absorb retroactive, class-action settlements as a licensing fee, the system is fundamentally broken. Paying a fine after the fact turns illegal data theft into a mere "cost of doing business."
Technology exists to do this ethically. Cryptographic provenance (like C2PA), data cloaking tools (like Glaze), and mandatory "opt-in" licensing frameworks aren't science fiction—they are available mechanisms that can (seemingly) protect active copyrights before ingestion. The argument from developers that it is "too hard" to separate copyrighted material from the public domain appears an excuse, not a technical limitation. Someone who knows more than me, chime in. Do these safeguards introduce more complicated dangers than we've considered?
I am putting this out there because I want to move past the binary, toxic arguments of "AI is purely good" or "AI is purely theft."
We need to navigate the grey area responsibly, we aren't escaping it.
This project began as an effort to create a quick referential image, testing one of many ideas for a self-portrait I plan to paint.
It's become a digital exploration of authorship, shaped by the historical weight of artists, driven by an algorithmic engine, and meticulously refined by the human hand.
I asked the AI model to reinterpret a selfie through the stylistic lens of Gustav Klimt. The result was version one. What followed was a deep, iterative collaboration with generative AI through almost two dozen versions. Through multiple branches and unexpected results. More specific tweaks (and rollbacks and try-something-differents). Some massive changes. carefully adjusting each element iconographically until it was just what I wanted.
I decided to record the whole process because somewhere I realized: I have more questions than answers.
Did I create the image? Did Google? Did we do it together?
If I paint it next, did I copy Nano Banana's art?
Does AI have a place in the arts? When is it an artist's tool, their collaborator, or their enemy? When does an artist fear that Generative AI will replace their job, and when do they incorporate it as a tool used in their regular creative process?
Let's Talk in the Grey Areas
I want to hear from the artists using AI and those losing to it. The art historians, graphic designers, marketers.
I want to hear from the experts building and regulating this frontier. To the IP attorneys, the machine learning engineers, and the digital rights advocates: How do we transition from a system of retroactive penalties to one of proactive, ethical licensing? How do we build an ecosystem where human artists and historical legacies are respected and compensated before the model trains?
I built Draftmode Co. as a public research project to explore this and other friction-filled questions we find ourselves facing today.
As I've begun the newest chapter of continued education into collaborative AI, I embark on this journey anchored with 30 years of experience as an artist, a foundation of art history, and over a decade navigating the technical side of the photography industry.
I am here to share what I know, research what I don't, and invite you to learn with me.
Is the cultural value of art diminished when AI is used for generation, even if intentionally conceived and heavily refined by the human artist creating it?
Yes - AI generated images can never hold any cultural value
Yes - Cultural value drops in correlation with human control
Potentially - Precise, intentional, novel art may hold value
No - Meaningful human control sufficiently instills value