The ethical questions surrounding AI-generated art are real, and they are not settled. Anyone working seriously in this space should be engaging with them — not because there are clean answers, but because the act of asking them shapes how you work, what you make, and what you are willing to put your name on.
This is not a verdict. It is an invitation to think carefully.
The Training Data Problem
Most AI image models are trained on datasets assembled by scraping the internet — billions of images that include the work of living artists, photographers, and illustrators who did not consent to their work being used this way. This is the most legally contested and ethically contested aspect of current AI image generation.
The industry position is that training on publicly accessible data constitutes fair use, similar to a human artist studying thousands of paintings to develop their style. The counterposition, advanced by many artists and increasingly by courts, is that the scale, automation, and commercial exploitation of this data constitutes something categorically different from human learning.
Both positions have force. The honest response, as an AI artist, is to hold this tension with honesty rather than resolving it prematurely. It is not sufficient to simply say "it's legal" — legal and ethical have never been identical. Nor is it accurate to say that every use of AI art tools is equivalent to theft.
Attribution and Transparency
If you are exhibiting or selling AI-generated work, disclosure matters. The question of whether an image was made with significant AI generation versus traditional methods is not merely academic — it affects how audiences understand and value the work, and it affects other artists' livelihoods.
The practice I advocate: be transparent, always. Label AI-assisted work as such. This is not a diminishment. The work in this exhibition is made with AI tools, and that is stated plainly. The transparency is part of the honesty of the project — the exhibition is, in part, about AI, so obscuring AI's role would undermine everything it is trying to do.
In commercial contexts — stock imagery, advertising, editorial illustration — the stakes are higher. Undisclosed AI generation in professional creative contexts is displacing work from human practitioners. That displacement deserves acknowledgement, not denial.
Cultural Appropriation and Representation
AI models trained predominantly on Western internet data carry the biases and representational defaults of that data. They tend to produce certain kinds of faces, certain kinds of bodies, certain cultural aesthetics with ease — and struggle with or distort others. They can replicate the surface aesthetics of non-Western cultural traditions without any grounding in the meaning or context of those traditions.
For an artist like me, working explicitly with Afrocentric and Afro-diasporic visual references, this requires active resistance. It requires knowing the cultural traditions you are working with deeply enough to recognise when the AI has flattened, distorted, or exoticised them. It requires correction, iteration, and sometimes simply not using AI for certain kinds of work.
The technology reflects the culture of those who built it and the data it was trained on. Engaging with it ethically means being aware of what it defaults toward and what it gets wrong.
The Question of Originality
There is a distinct but related question about originality and value. If a model has been trained on the work of thousands of artists, and you write a prompt that produces something that closely resembles a specific artist's distinctive style, have you created something original? Have you taken something from that artist?
The case for harm is strongest when specific living artists' styles are named in prompts to produce commercial competitors to their own work. The case for artistic freedom is strongest when working in general traditions, movements, or aesthetics rather than targeting individuals.
My own practice: I do not name living artists in prompts. I work with broad visual languages, cultural aesthetics, and historical references. This is a choice, not a legal obligation. But it is a choice made with awareness of who is on the other side of the question.
Asking Better Questions
The ethics of AI art are not resolved by avoiding the tools or by uncritically embracing them. They are navigated, imperfectly and continuously, by asking honest questions and letting the answers shape your practice.
What am I using this tool for? Who made the work it learned from? Am I displacing human practitioners who would otherwise be paid for this work? Am I representing cultural traditions I have a right to represent? Am I being transparent with my audience?
None of these questions have permanent answers. Technology changes, law changes, norms change. But the practice of asking them — and taking the answers seriously — is what distinguishes an ethical practice from an unconscious one.