With a mop of darkish curls and brown eyes, Jarren stands within the thick of the Australian outback, pink grime at his toes, a snake unfurling in entrance of him.
In a collection of on-line movies, the social media star, recognized on-line because the Bush Legend, walks via dense forests or drives alongside abandoned roads on the hunt for wedge-tailed eagles. Lots of the movies are set to pulsating percussion devices and yidakis (didgeridoo).
His voice appears like a cross between Gardening Australia’s Costa Georgiadis and Steve Irwin. His speech is peppered with “mate” and “crikey” as he passionately shares snippets with 1000’s of his followers about Australian wildlife – from venomous snakes and crocodiles to redback spiders and the elusive night time parrots, as soon as thought extinct.
His followers go away admiring feedback, marvelling at how he can get so near the animals and even suggesting he wants his personal TV present.
However none of it’s actual. The wildlife and the person presenting them are all creations of AI.
Created in October 2025, Meta signifies the account is predicated in New Zealand, with the Instagram account initially sharing an AI-generated satirical information account known as ‘Nek Minute Information earlier than pivoting to wildlife content material. Earlier incarnations of Bush Legend present the character carrying white physique paint seeming to imitate ochre, a beaded necklace adorning his neck.
As of this week, the Bush Legend account has 90,000 followers on Instagram and 96,000 on Fb. It says its focus is on constructing consciousness and training about Australian wildlife.
Guardian Australia has contacted the individual believed to have created the account, who’s a South African dwelling in New Zealand. They didn’t reply to a number of approaches.
‘Cultural flattening’
The selection to create an avatar of an Indigenous individual has raised moral considerations.
Dr Terri Janke, a lawyer and Indigenous cultural and mental property, knowledgeable, says the photographs and content material are “exceptional” of their realism.
“You suppose it’s actual, I used to be simply scrolling via and I used to be like, ‘How come I’ve by no means heard of this man?’ He’s lethal, he ought to have his personal present,” she says. “Is he the Black Steve Irwin? In his greens or the khakis, he’s a bit like Steve Irwin meets David Attenborough.
However whereas the Wuthathi, Yadhaigana and Meriam lady says the participating movies are “fairly unimaginable whenever you have a look at it as a device for training”, the creation of a seemingly Indigenous avatar is offensive and carries a danger of “cultural flattening”.
“Whose private picture did they use to make this individual? Did they convey collectively folks?” she asks. “I really feel a bit misled by all of it.”
AI-generated content material poses a specific danger to marginalised communities and could possibly be thought-about theft of cultural and mental property. It additionally probably takes alternatives away from genuine accounts, similar to movies created by the huge community of Aboriginal rangers.
“It’s theft that may be very insidious in that it additionally entails a cultural hurt,” Janke says. “Due to the discrimination … the impacts of stereotypes and unfavorable considering, these impacts do hit more durable.”
Janke says it’s potential to ethically use AI expertise to create content material about First Nations folks, nevertheless it requires the consent and involvement of First Nations folks.
Tamika Worrell, a senior lecturer in important Indigenous research at Macquarie College, says the AI avatar is a type of cultural appropriation and “digital blackface” the place a non-Black individual creates a Black or Indigenous caricature on-line.
The Kamilaroi lady says the proliferation of AI instruments with out acceptable legislative guardrails signifies that photos, cultural data and tales may be transmitted with out acceptable consent.
“AI turns into this new platform that we’ve no management or no say in it,” she says.
“Not solely tales or language however precise visuals of us can usually be taken from folks that have handed away – or simply mixing a spread of various folks [to create an AI avatar] with no form of accountability to the communities that these persons are from.
“It’s AI blackface – folks can simply generate artworks, generate folks, [but] they don’t seem to be really participating with Indigenous folks.”
The potential for hurt is twofold: such accounts default to sharing the “palatable” or “snug” elements of Indigenous cultural data and expertise, reasonably than the extra complicated actuality; and it additionally has the potential to amplify racism.
“I used to be wanting on the feedback from a contemporary publish of Bush Legend. We see the identical racist feedback that we all know mob on-line get. We see it once more utilized to an AI individual as nicely,” she says.
Toby Walsh, laureate fellow and scientia professor of synthetic intelligence on the College of New South Wales, says AI is skilled to breed info and rendering via large-scale information units with inbuilt biases, that means it’s not proof against racist or prejudicial content material.
“They’re going to carry the biases of that coaching information,” he says. “Sure teams could also be stereotyped as a result of the video information or the picture information that exists in that group on-line is considerably stereotypical. So we’re going to perpetuate that stereotype shifting forwards.”
Guardian Australia tried to contact the web page’s creator via a number of social media accounts and e-mail however was but to obtain a response.
The Bush Legend account has seemingly addressed the criticism via its avatar, saying the web page doesn’t search to “signify any tradition or group”. “This channel is solely about animal tales,” the AI creation stated in a video final week.
It went on to say the web page “isn’t asking for cash, donations or help” and content material is “free to look at”, and suggests folks “scroll on” in the event that they don’t prefer it. Earlier plugs requested followers to subscribe for $2.99 per 30 days.
Meta has been contacted for additional remark.
Walsh says whereas digital literacy can help customers with figuring out AI content material, the “tells” that assist determine it are getting more and more arduous to identify. “If not now, within the very close to future, it’s going to be subsequent to unattainable to have the ability to determine for your self whether or not this was actual or faux,” Walsh says.
“We used to imagine in issues that we see, as a result of it was once the issues that you just noticed have been largely actual issues.
“Now it’s not arduous to faux stuff. It’s extremely straightforward to faux stuff in a really convincing means, so we’re going to stretch the boundaries of what’s true and false.”
