‘Synthetic twin’: Why Socceroo Cristian Volpato is the latest footballer to license his AI clone

The Socceroos star looks at the camera as he walks through a huge hall. The sound of his football boots echoes across the polished concrete floor. Suddenly, four digital copies of him appear. They walk out of view and then appear again in different places around the world.

This scene in a social media video is fake and was made using artificial intelligence (AI). But the 22-year-old footballer, Cristian Volpato, has given permission for it to be made. He is the latest professional athlete to allow a company to use his AI-generated likeness in return for money.

In one scene, Volpato's “synthetic twin” wears a white linen suit and sits with his feet in the sand. He turns towards a photographer, with the limestone rock formations of the Twelve Apostles behind him. In another scene, a digital version of him wearing a football shirt fills a large digital billboard. A third smiles at himself in a mirror while adjusting a smart grey suit.

A new company that helps professional athletes license their AI likeness made the video. Experts believe this is a small part of a new online industry that could change how image rights and sports content work in the future.

Bookmakers, broadcasters and brands are increasingly using AI to create content featuring athletes without their permission. Some players are now fighting back.

Digital twins

In mid-September, Volpato, who plays for the Italian Serie A club Sassuolo, signed a deal with Callandor, a company based in California, allowing it to license his AI-generated likeness.

Under the deal, Callandor received exclusive rights to his digital likeness, voice, biometric profiles and performance data. The company will make money from his AI image by finding deals with brands and taking a share of the earnings. It will also help limit content that Volpato has not approved.

From Los Angeles, Callandor co-founder Michael Fisk tells Guardian Australia that his business creates “synthetic twins”.

“Cristian is constantly training,” Fisk says. “He's constantly out there playing. If he has any free time whatsoever, he'll probably want to spend it with family and friends, so he loses out on a lot of opportunities to appear in brand promotions.

“A synthetic twin could be used by brands 24 hours a day, all year round. Assuming it's approved, right?”

Callandor was founded in February and has a small group of athletes, including Brazilian footballer Jorginho and American sailor Chris Poole. Its advertising campaigns are still being tested, but the company is looking at many other ways to use AI beyond traditional advertising.

Fisk, who used to be an executive at Sony Pictures and MGM, talks about making AI-generated voiceovers for books, including versions in other languages. His team is also planning apps that would let fans have a personal connection with a digital version of an athlete.

“Ultimately, the holy grail of an athlete is to have that one-on-one interaction with fans,” Fisk says.

He suggests creating an app that watches how people play football. A digital copy of Volpato could then give them instructions as they play.

“He's realising I'm not handling the ball well, so his synthetic twin can tell me, ‘Hey, you should be doing this and this and this’,” Fisk says.

“Volpato would never be able to do that one-on-one with them.”

Volpato's deal with Callandor is at the leading edge of agreements about AI rights, but he is not the first athlete to use the technology. In June, Brazilian superstar Neymar allowed a Chinese-owned entertainment platform to use his AI likeness. The platform created short videos that were published during the FIFA World Cup.

These projects are starting as governments around the world consider basic rules for AI. At present, contracts for players in Australia's major sporting leagues do not mention AI.

But technology is moving faster than the rules, and Fisk says the digital twins his company works with are making it harder to tell what is real and what is fake.

“It's becoming almost impossible to detect the difference in look and feel, particularly when you see it on a mobile device,” he says.

‘We're watching a race’

William Saunders leads Hyperknown, a small start-up based in Sydney that is trying to make similar deals. He says athletes are starting to understand that if they do not make money from their AI likeness, other people may do so without their permission.

Earlier this month, the AFL told its gambling partner Sportsbet to stop using AI images of players without their consent. The move followed a complaint from the players' union, which is considering whether to negotiate stricter rules.

“Sportsbet shows us what happens when companies and technology get ahead of consent,” Saunders says.

“Volpato represents the counter-move: people establishing their rights before others start making decisions about their identities.

“We're watching a race between what technology can do and what people have actually authorised.”

Many people in Australia's sports industry, including content creators, bookmakers, broadcasters and athletes, have started using the technology.

Earlier this month, Australian cricketer Travis Head appeared in an advertisement for broadcaster Kayo. He joked that his AI clone had already finished filming the ad, before showing a video of himself riding a flying kookaburra.

Brand deals and endorsements have become major sources of income for athletes, with some earning millions of dollars from contracts. It is still unclear whether brands will pay as much to use an AI clone as they would to work with the real athlete.

But Saunders, whose company has only just started operating, says the opportunities are huge. He describes how AI clones could sell products in other languages in countries that the athletes have never visited.

“The next trust signal won't just say ‘made with AI’,” he says, referring to a label often used by social media companies. “It will say ‘made with this person's permission’.

“AI can make content abundant. It can't manufacture someone's permission. That's becoming a new commercial requirement.”

‘A tangle of different interests’

Jonathan Harlen, an expert in intellectual property law at Southern Cross University, says the rise of companies that manage AI rights is “the obvious next step” in a fast-growing global market.

From a legal point of view, he says, this new industry involves “a tangle of different interests and rights” that must be carefully negotiated.

“While copyright already exists in voice recordings, artistic works and film images, it is only recognised where a human has created the work and has put in some independent or creative effort,” Harlen says.

“It doesn't apply to AI reproductions.”

Celebrities and athletes have some legal protection against AI-generated content that uses their image and voice without permission. But laws against falsely suggesting that a person supports or is connected to something only apply when public figures can prove they have suffered a financial loss.

This has not stopped the rapid growth of deepfakes made with increasingly advanced software that anyone can buy and use.

Last month in the UK, some of the world's top footballers, including Cole Palmer and Lucy Bronze, took part in a campaign raising concerns about deepfakes and the failure of regulations to stop them spreading. The campaign was supported by Fifpro, the global union for professional footballers.

Earlier this year, an offshore casino used a deepfake video of Manchester United player Bruno Fernandes in an advertisement.

However, not every unauthorised deepfake is unwanted. Fisk says a strict crackdown on AI-made “fan art” could do more harm than good.

“You don't want fans to use your image and put you in content using brands that you don't use, or saying something that you would never say,” he says.

“Obviously, anything with religion and drugs, or whatever it might be, you have to be very, very careful. So it does complicate things a bit, but the opportunities are phenomenal.”

But in this new and developing area, it may be difficult to predict what will happen next.