Why AI Deepfakes Are the Next NIL Battle in Sports

The New Front in Athlete Endorsements

For decades, an athlete endorsement meant one thing. A brand reached out, the athlete signed a deal, the agency ran a campaign, and the athlete cashed a check for the use of their name and image. Every commercial appearance flowed through a contract the team could track, enforce, and renew.

AI is testing that model. Fans scrolling through their feeds now see videos of their favorite players hawking sportsbooks, supplements, financial products, and political messages the athletes had nothing to do with. The clips look polished enough that most people would never think to question them, and that is the problem. The endorsements never happened, but the content keeps spreading anyway.

None of this content goes through a contract. None of it pays the athlete. Most of it spreads faster than any team, agent, or compliance office can react. The pattern shows up across professional and college sports, where bad actors use athlete likeness to drive engagement and make money in places that sit completely outside the permission system the industry spent years putting together.

That permission system has a name. Name, Image, and Likeness rights, known as NIL, gave athletes their first real licensing setup. Generative AI now creates uses of that identity at a speed and volume the licensing setup was never built to handle.

This hits everyone from Premier League stars to college sophomores, and the athletes with the least support and the smallest legal teams take the hardest hits.

How NIL Built a Permission Layer for Athlete Identity

To understand why AI is such a problem for athletes today, it helps to look at how that permission system came together in the first place. It is newer than most people realize, and the legal scaffolding underneath it is still being built.

When the NCAA lifted its NIL restrictions in July 2021, athletes could finally sign endorsement deals, license their image, and earn from their personal brand. That moment ended a long stretch where schools, broadcasters, and game publishers built businesses around athletes who were barred from earning anything beyond a scholarship. The Ed O’Bannon case had already established right of publicity as a tool college athletes could use, and the 2021 rule change turned that legal foundation into actual market access.

State laws built on that foundation, with rules differing a lot from one state to the next. Some states limited certain product categories. Others required schools to help with disclosures. The result was a patchwork, but a working one.

The House v. NCAA settlement, approved in June 2025, pushed that system further. Division I schools can now share up to $20.5 million per school directly with athletes during the 2025-26 academic year, with the cap rising each year over a ten-year period. The settlement also includes $2.8 billion in back payments for athletes who competed between 2016 and 2024.

What this built, at its core, is a permission layer. Contracts spell out how athlete identity can be used. Money flows back to the athlete. Disclosure rules keep track of activity. College athletes now work inside a licensing setup that looks a lot like what professionals have had for decades.

This fits a broader pattern we have written about before, where identity is becoming licensed infrastructure across sports, media, and entertainment. NIL is one of the clearest examples of that shift inside sports. Then generative AI showed up and made the structure easy to route around.

AI Deepfakes and NIL Operate on Different Rules

NIL contracts cover specific, defined uses. A campaign, a social post, an appearance. Each use is made by a human team, placed in a known channel, and tied to a tracked agreement. Generative AI takes that bottleneck away. New uses can be made on demand, at scale, including AI deepfakes that look like endorsements but never touch a contract.

The clearest commercial example comes from international soccer, where scammers have used AI to make convincing fakes of athletes promoting financial products and betting apps they had no relationship with. Brazilian player Vinícius Júnior’s legal team has pursued similar cases, and Meta’s own Oversight Board overturned the company’s decision to leave up an AI-generated video of Brazilian soccer legend Ronaldo Nazário appearing to endorse an online gambling game.

The problem reaches well beyond elite players. Fake quotes, made-up charity gestures, and invented product ties involving college and professional athletes have spread widely, and some have been picked up by mainstream sports media before being flagged as AI-made. British regulators reported roughly eight million deepfake incidents across the country in the past year, with online betting platforms among the hardest hit.

The bigger issue is that NIL was built for a world where each commercial use of an athlete’s likeness had to be made and placed by a human team. AI takes that limit away. A licensing system built around one-off, negotiated uses now has to handle a world where uses can be made faster than anyone can check them.

The fallout lands hardest on athletes with the least support.

Why College Athletes Face the Greatest Deepfake Exposure

College athletes sit at the sharpest end of this problem, and three factors stack up to put them at greater risk than professionals. Each one is workable on its own. Together, they leave most college athletes exposed in ways the NIL system never planned for.

1. The Support Gap

Most college athletes do not have full-time agents, legal teams, or brand managers checking what is being shared about them online. A pro athlete might have a management company handling brand monitoring as a paid service, while a college sophomore is often running their own social presence between class and practice. The work that gets absorbed by professional teams falls directly on the athlete, who usually has no setup for it.

That support gap matters more when the content itself can carry real penalties.

2. Eligibility Risk From Unauthorized AI Content

A viral AI-made endorsement of a sportsbook, supplement, or political figure can damage an athlete’s existing NIL deals. Depending on the state, the school, and the conference, it can also affect eligibility, even when the athlete had no part in making the content. The job of proving the content is fake falls on the athlete, who often learns about it only after it has already spread.

Where an athlete plays also shapes how much they can do about it.

3. The Legal Patchwork Across States

Tennessee passed the ELVIS Act in 2024, becoming the first state to specifically protect voice and likeness from unauthorized AI use. As of spring 2026, 46 states have passed some form of deepfake law, with the federal TAKE IT DOWN Act setting a baseline for intimate imagery. States including Utah, Arkansas, California, Virginia, and New Jersey have passed laws specifically aimed at AI-generated content targeting student athletes. Athletes in states without those protections have fewer options, and content crosses state lines no matter where any one law applies.

On top of all this, college athletes are building their personal brand in real time. A fake piece of content during a recruiting window, a transfer portal period, or a championship run can affect deals that have not even been signed yet. The damage often shows up in opportunities that never materialize rather than in deals that get formally lost.

Why Athletes Cannot Detect Deepfakes Alone

Most of this damage happens out of view of the athlete, which points to a deeper issue. Even where legal protections exist, finding unauthorized content in the first place is its own problem. The real issue is less a deepfake problem than an identity problem, because anyone with a short clip or a single image online can have their likeness reused, no matter how big their following is.

For college athletes, the daily math makes manual checking unrealistic. Class, practice, travel, recovery, and endorsement work fill the day. Searching across TikTok, Instagram, X, Reddit, YouTube, betting forums, and overseas platforms is not something most athletes can do on top of that. Even when an athlete finds unauthorized content, takedown processes vary by platform, often need detailed paperwork, and usually move slower than the content itself.

Platforms have also been all over the map on enforcement. Internal documents reported by Reuters in late 2025 suggested that roughly ten percent of Meta’s 2024 advertising revenue came from ads tied to fraud operations and banned products. The Oversight Board has noted that at-scale content reviewers often do not have the authority to act on impersonation content that uses a famous person’s likeness for fraud, even when the content has clear signs of breaking the rules.

The result is that most unauthorized AI uses of athlete likeness are never reported, never taken down, and never paid for. The question shifts from whether athletes have legal protections to whether the system exists to actually enforce them at the scale AI now works at.

What Proactive Deepfake Protection for Athletes Looks Like

A working system for athlete identity protection would need to do four things on the athlete’s behalf, and each one fills a different gap in how today’s setup handles unauthorized uses.

It starts with a verified baseline. Before any system can flag unauthorized content, there has to be an official version of the athlete’s identity that the system can check against.

1. Registration of Athlete Likeness and Voice

An athlete’s likeness, voice, and biometric markers need a place where they can be formally registered as protected. Athlete AI registries are starting to show up in professional sports, with similar setups appearing in entertainment through services like Creative Artists Agency’s CAA Vault, which stores digital replicas of clients’ faces, bodies, and voices under controlled access.

Registration is only useful if something is actively looking for matches. That brings the second piece into play.

2. Automated Detection Across Platforms

Scanning across platforms for unauthorized use of registered identity is where the heavy lifting happens. The closest thing we have today is YouTube’s Content ID system for music, but applied to faces, voices, and movement patterns rather than audio fingerprints. YouTube has begun opening its likeness detection tools to actors, athletes, creators, and musicians, including those without active accounts on the platform. The technical foundations are there. The use for athlete identity is still early.

Detection on its own still leaves audiences guessing about what is real. The third piece closes that gap.

3. Content Verification for Audiences

Viewers need a way to confirm whether a piece of content actually came from the athlete. Browser extensions and content authentication services have started to appear that let viewers check whether a piece of content has been verified by the person it shows. These work best when paired with a registry that sets a clear baseline for what real content looks like.

Spotting unauthorized content matters only if something happens after it is found. That is where the system has to deliver real consequences.

4. Enforcement and Takedown at Scale

Takedown processes have to act on the athlete’s behalf without making them file each one personally. Managed services that search the web and dark web for deepfake content and trigger legal action are starting to appear for elite athletes, but the cost structure right now limits access to athletes who can afford private security firms. For the system to mean anything at the college level, enforcement needs to be available without retainer-level fees.

The pieces exist in fragments across different vendors and platforms. What does not yet exist is a connected system that ties detection and enforcement back to the NIL permission layer, so unauthorized uses can be found, taken down, and where it makes sense turned into licensed ones.

The Future of NIL Rights in an AI World

Building that connected system is the work ahead, and it changes what NIL itself has to mean. NIL gave athletes a contract layer. AI is making clear that a contract layer alone does not protect anyone when no one is watching the platforms where unauthorized uses appear.

There is a longer-term question about how this changes athlete identity itself. If unauthorized AI uses become common enough, the value of any single NIL deal becomes harder to size up, because the athlete’s likeness is already moving through channels neither party controls. Brands looking at deals will start weighing detection power alongside reach and engagement. Athletes weighing offers will start asking what protection comes with the contract. The market for athlete identity will start pricing in the cost of policing it.

The next layer of identity infrastructure for athletes has to be proactive. It has to find unauthorized uses before the athlete does, because in most cases the athlete never will. It has to work across platforms and across state lines, because that is where AI-made content moves. And it has to tie detection back to the permission system, so unauthorized uses can be found, taken down, and tracked against the same licensing setup that governs the real ones.

Identity protection for athletes is moving from a contract problem to an infrastructure problem.

FAQs