People online duped for thinking AI influencer Mia Zelu is real as deepfake accounts skyrocket across social media

News.com.au

22-07-2025

Influencer Mia Zelu's done it all.
Courtside among the biggest names at Wimbledon? Yep. A Coldplay concert experience of a lifetime? Completed it. Sipping coffee in picturesque Italian streets? Piece of cake.
Her Instagram page is filled with big bucket list stuff. Except it's not real.
Not the typical social media personality, Zelu is actually an AI influencer, meaning she's just a generated picture on a screen. Nothing more.
But she looks so real that most of her 167k Instagram followers wouldn't even know she doesn't actually exist.
Don't tell those who keep up with her 'sister' Ana Zelu, who's fooled even more people with 267k followers who interact with her either unknowing or uncaring of the truth behind her account.
With their photorealistic posts and human-like captions, the fake sisters are just a few of the increasing number of AI accounts that are fooling people into thinking they're real, despite (some of) their bios stating they aren't.
Tech expert and editor of EFTM.com Trevor Long says the reason these accounts are having the same effect on people as real influencers is because AI has understood what people are drawn to and can feed into the same patterns without skipping a beat.
For a technological tool designed to help people, having it understand what people want isn't a bad thing. The danger is not knowing what's real and what's not.
'Most of us don't know the influencer on the other side of [an] Instagram account that is real, so knowing that someone is real or artificial intelligence actually doesn't change much of the perception of content,' Mr Long told news.com.au.
'However, if that content is sculpted and created in such a way that it is truly targeted and you don't have the morals of a real human being deciding whether or not they will sit in that spot, take that photo, try that thing, go to that event, we start to really push the boundaries of where this influential culture might go.'
Getty Images' Asia-Pacific head of creative Kate Roruke said they've conducted research that found that although 65 per cent of people could spot an AI photo, more than 95 per cent also mistook real images for AI.
'People are used to seeing curated, almost perfect images from human influencers, achieved through extensive editing, filters and professional photography. Zelu, being entirely AI-generated, naturally embodies this idealised flawless skin, perfect lighting and picture-perfect poses,' she said.
But concerns then about the extreme uses of the tech then also create a problem, like deep fake pornographic material which has pushed the moral and ethical boundaries of AI.
Numerous celebrities like Taylor Swift and face of the NRLW Jaime Chapman have already become victims.
The value of knowing if something is real has never been more important in an age when a tool not everyone yet understands is already out of control.
Mr Long says it is incumbent on the big tech companies like Meta and TikTok to be able to give users validation on what is real and what isn't, and give precedence to the real people using their platforms.
'We talk so much about the algorithm. It should be the case that real people are prioritised so that we know that we can listen to and decide whether or not we trust that person, otherwise we're probably putting our trust in an AI fake individual,' he said.
While easier said than done, companies like YouTube have taken steps towards creating better clarity and priority to real content, last week announcing they were demonetising accounts and channels that generate purely AI generated content.
'There's some fun AI videos out there. It might be a kangaroo doing a vlog or silly things like that,' Mr Long said.