If you have noticed a trend going around of “magical” avatars floating around on social media you wouldn’t be alone.
The AI-generated images are allowing people to see how they might look in a fantasy world or perhaps a real world from centuries ago.
So, how does it work and why do some people say there might be issues with the technology?
Where are these images coming from?
Users upload a number of headshots onto an app called Lensa — an AI generator owned by US-based company Prisma Labs.
From there, the application uses a technology called stable diffusion — a technology in which the AI has been trained with billions of images and mixes them together to create new ones.
Users upload their photos onto the app and within 10 or so minutes it generates 50 images in five different artistic styles.
Loading Instagram content
The cost of the images has been fluctuating due to increasing popularity, but currently in Australia users can pay $4.49 a week, $7.99 a month or $53.99 for a year, however free trials are available.
The app has been around since 2018. However, the addition of the “magic avatar” feature last month saw millions download the app and join the trend.
A number of celebrities, including Chance the Rapper, have also put their selfies through the app and shared the results online.
Why are artists upset?
Lensa learns from art that already exists, which some people on social media have called stealing.
Some have even pointed out that the images have scribbles similar to signatures on them.
Australian National University intellectual property expert Surend Dayal told ABC Radio there was not much recourse for artists concerned their work was being used to generate profits for apps like Lensa and their owners.
“[The artists] definitely have ownership In the original work. The question is whether they have any claim of ownership in the AI-generated work,” he said.
“At the moment, it would be difficult to contend that.”
What does the company say about it?
In a Twitter thread, the company defended the AI generating similar artistic styles to humans.
“AI is capable of rapidly analysing and learning from large sets of data, but it does not have the same level of attention and appreciation for art as a human being,” Prisma said.
“AI produces unique images based on the principles derived from data, but it can’t ideate and imagine things on its own.
“As cinema didn’t kill theatre and accounting software hasn’t eradicated the profession, AI won’t replace artists but can become a great assisting tool.”
Should I be concerned about uploading images of myself?
The app asks users to upload 10 to 20 images of themselves so it can create artwork in these styles.
However, in the past other applications such as FaceApp — which generated images of people of different genders and ages — came under scrutiny for the amount of personal information the Russia-based company collected.
Katherine Manstead, director of cyber intelligence and public policy at CyberCX, told ABC Radio there were often nefarious reasons companies put these apps out.
“One [of the reasons] is they’re trying to train their AI models so they can make more money down the track,” she said.
“If you look at Lensa’s terms of service, they’re pretty explicit that they will use your images that you create through the app to train their AI to make it better.”
She said the images it asks for provide the company with powerful biometric identifiers.
“We should think really carefully before we hand it over,” she said.
“We do know — and there are examples in the past regardless of where the company is located — that companies with powerful facial recognition models and powerful artificial intelligence models, they’ll sell data.
“They’ll sell their data to companies, to governments, to law enforcement.”
Lensa says once the avatar is generated, the company deletes the images, but the terms state once you upload your photos or “User Content” you grant it “perpetual, irrevocable, non-exclusive, royalty-free, transferable, sub-licensable license to use, reproduce, modify, create derivative works of your User Content.”
Are there any other issues?
Some users have called out Lensa for over-sexualising female users and for “white-washing” people of colour.
One person even told WIRED the app created child sex abuse material style images, along with other nude images despite the “no nudes” and “adults only” policy.
CEO Andrey Usoltsev told TechCrunch the company was aware of the issue but said the nude images were only generated when the AI was “provoked”.