NIL at Five: AI Is the Next Battleground for Athlete Likeness Rights

Name, image and likeness compensation for college athletes just passed its fifth anniversary, and the ground has shifted again. The House v. NCAA settlement, approved in June 2025, let schools pay athletes directly for the first time and built a new compliance apparatus around third-party deals. But the most interesting legal questions in the NIL space right now are not about revenue sharing caps. They are about artificial intelligence, which is simultaneously running the new enforcement machinery, threatening athletes with synthetic endorsements they never made and forcing a rethinking of what a likeness even is. Athletes are now small businesses, and like every business I advise, they have an AI problem whether they know it or not.

The New NIL Machinery

A quick refresher on where things stand. Under the House settlement, Division I schools may share revenue directly with athletes, with a first-year cap of approximately $20.5 million per institution that escalates over the settlement term. Third-party NIL deals of $600 or more must be reported and must serve a valid business purpose at a reasonable range of compensation. Oversight belongs to the new College Sports Commission, which reviews deals through its NIL Go platform. The system has teeth: NIL Go has cleared more than 17,000 deals and rejected more than 500, and rejected athletes can pursue neutral arbitration, as 18 Nebraska football players recently did over roughly $7.5 million in blocked multimedia rights deals. The Commission prevailed, and the message was clear. This is now a regulated market.

The Algorithm Is Already in the Room

Here is the first AI issue, and it is hiding in plain sight. NIL Go, built with Deloitte, evaluates whether a deal falls within a reasonable range of compensation by benchmarking it against market data. Whatever label you put on the underlying analytics, athletes’ economic opportunities are now being screened by automated valuation models. Meanwhile, AI-driven platforms are springing up to estimate athlete market value for schools, brands and families on the front end. This should sound familiar to anyone who follows algorithmic decision-making in lending, hiring or insurance. When a model determines that a deal exceeds fair market value and the deal dies, the athlete deserves to know what the model considered, whether the comparables were sound and how to challenge the output. The Nebraska arbitration was the first major stress test. It will not be the last, and counsel negotiating NIL deals should be building the valuation record they will need if a client’s deal gets flagged.

Deepfakes Are Coming for the Product Being Sold

The second AI issue is more visceral. The entire value of an NIL deal is authenticity, and generative AI has made authenticity forgeable. A convincing video of an athlete promoting a product can now be created without the athlete ever stepping in front of a camera, approving a script or signing anything. This is not hypothetical. AI-generated content falsely claimed that Patriots quarterback Drake Maye was donating his wedding gifts to charity, and major outlets reported it as fact. NHL star Brady Tkachuk appeared in an AI clip he never made. For a college athlete, the harm runs in every direction at once: fake endorsements erode the exclusivity that real sponsors pay for, synthetic merchandise siphons licensing revenue and a fabricated controversial statement can destroy a brand built over years. Unauthorized AI content also creates a compliance trap unique to college sports, because an athlete who appears to endorse a gambling site or supplement brand, even falsely, invites eligibility scrutiny before the truth catches up.

The legal toolkit exists but is fragmented. The right of publicity, the same doctrine Ed O’Bannon used against the NCAA over video game avatars, forbids commercial use of a person’s identity without consent, and Lanham Act false endorsement claims can reach AI-generated promotions. States are moving too. Florida updated its altered depictions statute with platform takedown deadlines aimed squarely at protecting its enormous population of NIL-earning athletes, Tennessee’s ELVIS Act extended protection to voice and New Jersey enacted its own deepfake liability law in 2025. The problem is that a deepfake posted from anywhere reaches everywhere, and a patchwork of state remedies is a poor match for that reality.

Washington Is Finally Moving

Congress noticed. The NO FAKES Act of 2026 would create a federal intellectual property right in a person’s voice and visual likeness, with liability for producing or distributing unauthorized digital replicas, a notice and takedown process and platform liability for knowing hosts. On June 18, 2026, the Senate Judiciary Committee advanced the bill unanimously, sending it to the full Senate. For athletes, a federal digital replica right would be transformative, replacing fifty inconsistent state regimes with a single national cause of action. Notably, the right would be licensable but not assignable during life, a structure with real consequences for how NIL agreements, collectives and agencies can package replica rights. Some athletes and celebrities have tried to build federal protection through trademark filings on elements of their persona, a strategy Matthew McConaughey and Taylor Swift have pursued, but trademark law was never designed for synthetic media, and NO FAKES would close that gap directly.

The Contract Is the First Line of Defense

Legislation takes time, and litigation takes longer. The immediate protection is drafting. Every NIL agreement signed in 2026 should answer a set of questions that did not exist when the form contracts were written. May the brand create a digital replica of the athlete, and if so, for what uses, in what media and for how long? May the athlete’s content be used to train AI models? Does the license terminate cleanly, with an obligation to destroy replica assets, when the deal ends? Who bears the cost of monitoring and takedowns when third parties generate fake content, and does the brand have approval rights over AI-assisted creative that features the athlete? Schools have parallel homework, since group licensing programs, media rights deals and even routine model releases now need to address digital doubles explicitly. A release drafted in 2019 that grants use of an athlete’s image “in any media now known or hereafter devised” reads very differently when the licensee owns a generative model.

NIL began as a fight over whether athletes could be paid for their identities. The next fight is over who controls the synthetic versions of those identities, how algorithms price them and what happens when machines manufacture endorsements no one gave. Athletes, schools, collectives and brands that get their contracts and compliance posture right now will be ready for it. The ones relying on 2021 paperwork will not.