Sally Scales, an artist and chair of the National Gallery of Australia's First Nations advisory group, has said generative AI is taking Aboriginal art and turning it into "Temu slop." According to Scales, the technology erodes artists' control over their own work and the authenticity of their creations; it amounts to the "theft" of cultural knowledge. The debate goes beyond the question of copyright infringement in AI art.
What happened?
A figure representing First Nations art has brought into the public arena the fact that image-generating models are learning the style of Indigenous art and flooding online marketplaces with cheap imitations. The claim rests on two points: artists are being stripped of control over their work, and cultural heritage is being used as training data without consent. That is the factual core; the rest is why it matters so much.
Why is this not just a copyright dispute?
The debate over AI and art mostly plays out in terms of individual copyright: a painter's style was learned without permission, an illustrator's work ended up in a dataset. With Aboriginal art, however, the issue sits at a deeper layer. Most of these works are not decorative; they are tied to specific communities, stories and knowledge of place. Who may draw a motif, and which story belongs to whom, is governed by rules internal to the community. So the question is not only "who is making money" but "who has the right to tell the story."
That is why the words Scales uses — "authenticity" and "consent" — are not technical but cultural. When a model digests thousands of Indigenous works and produces a statistical average, the resulting image looks "Aboriginal in style" on the surface, but the knowledge, connection and permission beneath it have been severed entirely. This is the reduction of cultural context to a visual shell.
Australia had developed mechanisms to combat forgery in Indigenous art: labels documenting provenance, community-approved sales channels, "genuine Indigenous work" certificates. These tools were designed to weed out physical fake paintings. They do not work against cultural heritage scraped as training data, because AI output is not a copy; it is a new image that resembles no single work exactly, yet is derived from all of them. The existing legal framework asks "did you copy the same thing," whereas the problem is at the level of "you extracted the style and the knowledge en masse."
Who does it affect, and how?
The economic impact is felt most sharply by Indigenous artists themselves. For many remote communities, art is one of the few sources of outward-facing income. This economy already runs on thin margins and depends on intermediaries. When the marketplace fills with cheap imitations, two things happen at once: price pressure and an erosion of trust. Buyers can no longer tell the real from the generated.
| Dimension | Traditional forgery | AI imitation |
|---|---|---|
| Production speed | Slow, handmade | Unlimited within minutes |
| Cost | High | Almost zero |
| Detection | Possible with provenance documentation | Hard, because it is not an exact copy |
| Legal basis | Copyright and labelling laws | Largely in a vacuum |
The difference is scale and speed. Hand-made forgery produced a limited number of works; automation does the same job on an industrial scale. That is why it seems more accurate to read this not as "a new copyright case" but as the automation of cultural exploitation.
What's next?
There are a few things to watch. The first is legal: calls for regulation recognising Indigenous cultural and intellectual property have existed in Australia for years, and AI will likely accelerate that demand. The second is the platform side: whether marketplaces will make an "AI-generated" label mandatory will be decisive, because the visible part of the problem lies in the sales channels. The third is the data side: transparency about which images models were trained on, and the exclusion of content scraped without community consent.
The realistic assessment is this: the existing tools were not designed for this problem, and the technology is advancing faster than regulation. The solution will most likely come not from a single law but from provenance verification, platform responsibility and data transparency built together. Until then, the party bearing the burden will be the artists most dependent on this very economy.
Yorumlar (0)
Henüz yorum yok. İlk yorumu sen yaz.
Yorum yapmak için Sinyal'i indir
Yorumlar Sinyal hesabıyla yapılır. Mobil uygulamada giriş yap, yorum bırak.