Cameras cover a professional athlete from every angle. The game, the tunnel walk, the postgame press conference, media day, the sponsor shoot back in June. Most of them then post plenty of their own on top of that. A few seasons in, there are thousands of clean, well-lit images of one person available to anyone with a browser.
That face is doing real work. It is why a brand signs them, why a kid picks their jersey, why a post from them lands differently than the same post from a stranger. A convincing fake borrows all of that without asking, and the athlete is usually the last person in the chain to find out.
None of which means every AI image of an athlete is a problem. Fans have been drawing, editing, and photoshopping the people they follow for as long as sports has existed, plenty of athletes enjoy it, and some are actively looking for ways to take part. The goal was never to keep athletes out of AI content. It is to give them a say in which uses they are fine with and which ones they are not.
Why Athletes Are Particularly Vulnerable to AI Deepfakes
Being faked is not unique to athletes. Musicians, actors, and people with no public profile at all are dealing with the same tools, usually one report at a time.
What makes sports different is that four things happen at once. The raw material is already sitting there in bulk, the audience is ready to believe what it sees, the news cycle moves too fast for anyone to stop and check, and the likeness being borrowed already has money attached to it. Any one of those on its own would be a small problem.
1. There is more material on them than almost anyone
Professional coverage is only half of it. Twenty thousand people in the stands are shooting the same person on their phones, and a good share of that is online before the final whistle. Almost nobody outside of sports is photographed by that many strangers in a single afternoon, which means the footage that makes an athlete marketable is also what makes them easy to replicate. An athlete does not have to post anything careless or have anything leak. The library builds up just by doing the job.
2. Fans recognize the face before they think to question the post
People recognize the face first and question the post later, if at all. C. Shawn Eib, who runs investigations at the AI risk firm Alethea, made the point to Front Office Sports that people have learned to be careful with political content. They have never had a reason to be careful with sports, which leaves the whole category open. Researchers who study misinformation keep finding the same thing. Whichever version of a story someone hears first is the one that sticks, and the correction rarely catches up.
3. Sports news already moves fast and comes from strangers
Trade deadline, injury report, contract talks, a locker room story nobody has confirmed. Fans are used to news breaking fast from accounts that are not official, so a fake post does not stand out. It looks like everything else in the feed. Alethea's report on NFL content in January found pages posing as team fan sites posting upward of 50 times a day, run from overseas, making their money on links out to ad-stuffed junk sites.
4. The likeness already has contracts wrapped around it
Most athletes at this level have sponsorship deals that limit which product categories they can appear in. So an AI post putting them in a sportsbook ad is not just embarrassing. It sits on top of an agreement somebody spent months negotiating, and the athlete has to get on the phone with a partner about something they had nothing to do with. This is not only a professional problem either. NIL means a college sophomore can have real deals attached to their face, usually with nobody behind them to sort out a fake when one shows up. Fake ads are the clearest example of all this, and we wrote about AI-generated endorsements on their own.
How an Athlete's Likeness Can Be Misused With AI
Fake ads are also where most coverage stops. There is no shortage of them, and they are only one of the things being made with an athlete's face. A lot of what turns up is not selling anything at all.
Fake statements are the most common type. After Team USA won hockey gold at the Milan Cortina Games in February, a video spread in which Brady Tkachuk appears to insult Canadian fans in language he never used. AFP traced it to an account that posts AI parody quotes of athletes. Tkachuk told reporters it was not his voice and that "I can't do anything about it." He is an Olympic gold medalist with an agency behind him, and that was still where it ended.
Other types include:
- Invented news. Trades, signings, injuries, retirements, coaching hires, all of it generated for engagement rather than accuracy. Alethea's investigators pointed out one easy sign, which is a single person being reported as hired by several teams at once.
- Gambling and fraud content. Meta's Oversight Board looked at an altered video promoting a gambling app, then searched Meta's own ad library. It found thousands of active ads for the same app, several using AI endorsements from famous footballers.
- Sexualized imagery. During the 2026 Winter Olympics, the research firm Graphika documented users generating and trading non-consensual intimate imagery of several women competing at the Games, made with downloadable AI models that run outside the limits built into mainstream tools. Women athletes carry more of this than anyone, though nothing about the method is limited to them.
- Fabricated footage and invented backstories. Euronews went back through the 2026 World Cup afterward and found AI clips of players, doctored photos, and staged evidence that had pulled in millions of views before anyone corrected a single one.
- Political association. Athletes keep getting written into election content they had no part in, which is a problem reaching well past politicians this year.
Protecting Athletes' Likeness at the Scale of Sports
Finding this content and getting it taken down is a job, and hardly any athlete is doing it themselves. There is usually an agent, a management company, a club, a communications team, or a player association involved, and any one of those can end up looking after a few hundred faces at once. At that scale, three problems show up straight away.
1. The work is nobody's job in practice
An agent is running deals, travel, media requests, and whatever went wrong that morning, and a communications team is busy with the actual season. Searching a handful of platforms for every unauthorized use of a face, checking whether it breaks anything, filing the report and chasing the reposts is a full job in itself. At most organizations it is nobody's. So it tends to get picked up only after something has spread far enough to cause a problem, by which point the sponsor has already seen it and the athlete is answering questions about it.
2. Every athlete draws a different line
The differences are specific. One player has a sportsbook exclusivity clause that turns any gambling association into a same-day problem. Another will not be shown in a rival club's colors under any circumstances. Another wants their face nowhere near a political position for reasons that are entirely their own. Another loves the fan edits and would be annoyed if anyone took them down.
Which is why noticing that a piece of AI content contains a particular athlete settles almost nothing. The real question is whose likeness it is, what the use is, and whether that use crosses the boundary that person set. Two nearly identical images of two different athletes can land on opposite sides of the line, and only the athletes know where their own line sits.
3. Where legal protection stops short
The law helps with some of this, but only after something exists, and only in some states. The right of publicity lets an athlete act when their likeness is used to sell something, and what it covers changes at the state line. None of it tells anyone in advance what a particular athlete would have said yes to, and plenty of them would say yes to more than people assume.
Protecting Athletes' Likeness Doesn't Mean Stopping AI Creation
Saying yes matters more in sports than in most places, because fans here make things constantly. Edits set to music, hype graphics, imagined matchups between eras, tribute videos cut together when someone retires. AI tools fit straight into that, and most of what comes out is made by people who like the athlete.
The 2026 World Cup showed how big it can get. Kylian Mbappé spent the tournament as the subject of an enormous running joke, turned by AI videos and fake posters into a ruler giving orders to managers, referees, and world football in general. Time Out counted him among the players who became meme royalty over the course of a few weeks. He had nothing to do with any of it. He did not start it, could not steer it, and had no way of saying which versions he was fine with and which he was not.
Plenty of athletes would enjoy being at the center of something like that. Others would see a version of themselves cast as a dictator and want it gone. Both reactions are fair, which is what makes this hard, and it is why no one setting works for a squad, let alone a league.
Where an athlete does want to take part, the upside runs past the licensing fee. Fan creation is engagement, and engagement is what an athlete's commercial value is built on. Every image a fan makes and posts puts the athlete in front of an audience they did not have to pay to reach, and that activity is exactly what a sponsor looks at when deciding what a deal is worth.
Fans get something out of it too, which is the certainty that what they made is welcome instead of a guess at where the line sits. And the athlete earns from images carrying their face rather than watching that value land somewhere else, which is closer to how AI generation works with permission already.
Block what they do not want, open up what they do, and let the athlete be the one drawing that line.
Giving Athletes More Control Over Their Likeness
All of it rests on the same thing. Being recognizable is what makes an athlete's likeness valuable, and it is the same quality that makes it worth faking. That trade has always been part of the job. AI just made one side of it close to free.
The two loudest answers do not work. Shutting AI out of sports was never going to happen, and telling fans to stop making things about the athletes they follow would cost the sport something real. What is actually missing is smaller than either. Athletes need somewhere to say, in advance, what their face can be attached to. They also need something that acts on it when a use crosses that line, rather than hearing about it from someone else.
That is what PersonaShield is built around. An athlete or their representative sets their safeguards once, and those safeguards drive the monitoring and takedowns on the protection side while deciding what fans can officially create, with the athlete earning from those images.
Protection and permission come out of the same decision, and it belongs to the person whose face it is. If you are an athlete or you represent one, get in touch and we will walk you through how it works.
