Indigenous Australian art carries something that most digital content does not: cultural knowledge encoded across generations, tied to specific communities, families, and lived experience. When AI systems scrape, alter, and reproduce that art without consent, the damage is not only economic. It is a form of cultural erasure. That is the argument being made by artists, legal scholars, and advocacy groups in Australia, and it is worth understanding precisely why this case is harder to resolve than standard copyright disputes.
What “Temu Slop” Actually Describes
Sally Scales, a Pitjantjatjara artist and chair of the National Gallery of Australia’s First Nations advisory group, has used a pointed phrase to describe what AI is producing: “Temu slop.” The reference is deliberate. Searches on platforms including Temu, Shein, and Amazon surface large numbers of items described as Aboriginal paintings or bearing Aboriginal-style designs, most without attribution to any artist. Coasters, stubby holders, decorative objects, all carrying visual patterns that draw from Indigenous traditions, generated or altered by AI, sold without connection to the communities whose knowledge they imitate.
Scales’s critique goes beyond aesthetics. The problem, as she frames it, is that the artist’s family lineage is absent. The lived experience is absent. The connection that gives the work its meaning is absent. What remains is a surface pattern, stripped of everything that made it significant. “It’s a theft of the agency of the artist,” she has said. That framing matters: this is not just about money or attribution. It is about who has the right to make something, and what is lost when that right is bypassed by a system that treats all online content as equivalent raw material.
The Legal Gap That Makes This Worse
Intellectual property law was not designed with Indigenous cultural knowledge in mind. Prof Kathy Bowrey, an intellectual property specialist from UNSW Law, has noted that one in five creators already faced copyright infringement problems before AI entered the picture. The addition of AI makes those problems more complex, and for First Nations creators specifically, the complexity is compounded by what she describes as a lack of control over ownership and respect for Aboriginal culture.
The mechanism she describes is instructive. An organisation pays to license artwork or images of Aboriginal people. Later, it uses AI to update or repurpose those images without returning to the original artist. The artist receives no repeat business. And when they create new work, they find themselves competing against an AI trained, in part, on their own output. The corporate client, in this scenario, has effectively purchased a one-time licence and converted it into an indefinite generative resource.
Terri Janke, a Wuthathi and Meriam lawyer recognised as an international authority on Indigenous cultural and intellectual property, has documented AI systems scraping Indigenous art centre websites, removing artists’ names, and altering their work. Janke has also raised concerns about AI reproducing cultural knowledge without regard to its accuracy or appropriate use. This is a distinct risk: sacred or ceremonially significant knowledge, embedded in visual work, being reproduced in contexts where it has no business appearing, by systems with no capacity to understand what they are handling.
The National Association for the Visual Arts (Nava) has submitted to Australia’s joint select committee inquiry on artificial intelligence that AI-generated images imitating community-specific techniques, stories, motifs, and mark-making are being sold as Indigenous or Indigenous-inspired work. The result, Nava argues, is misinformation alongside cultural and economic harm.
Why This Case Reveals a Structural Problem in AI Governance
The Australian government has stated there are no plans to weaken copyright protections, and the attorney general’s office has described ongoing consultation with creators, media organisations, and AI companies. But the gap between stated intent and current reality is visible. AI systems have already been trained on the work of well-known Australian artists without compensation, and the legal frameworks that might address this are still being debated.
Bowrey and her colleague Oliver Bown, a creative technologies expert, have proposed what they call a Copyright Plus framework. The proposal would require AI companies to disclose their data sources, give artists the right to opt out of having their work scraped or mimicked, impose a levy on AI companies to fund compensation for creative professionals, establish a cultural fund, and enable takedown notices for fake content. This is a policy proposal, not law, but it illustrates the kind of structural intervention that advocates believe is necessary.
Here is what most coverage of AI and copyright misses: the standard debate centres on economic compensation, and that matters. But for Indigenous cultural knowledge, the harm is not reducible to a royalty calculation. Cultural material carries obligations, protocols, and meanings that exist outside the market. An AI system cannot honour those obligations. It cannot know them. And current law provides no mechanism to enforce them.
In Short
AI systems trained on publicly available content are reproducing Indigenous Australian art without consent, attribution, or any understanding of the cultural knowledge embedded in that work. The economic harm is real: artists lose business and find themselves competing against systems trained on their own output. The cultural harm goes further: sacred knowledge, community-specific techniques, and family lineage are stripped away, leaving imitations that spread misinformation and dilute meaning. Existing copyright law was not built to address this. Proposed frameworks like Copyright Plus point toward possible remedies, but none are yet in place. The core issue is not technical. It is about who controls cultural knowledge, and whether legal systems can catch up before the damage becomes irreversible.
Based on reporting from The Guardian.