AI is entering training. Athlete data is not ordinary production data.

01The training ground is becoming a continuous data system

The NFL and AWS combine training, practice and game video with movement data in the Digital Athlete, running millions of match and scenario simulations. Every club can view its own daily loads and injury risk alongside league averages and historical benchmarks.

The system also informs rule design. When studying the Dynamic Kickoff, the NFL simulated 10,000 seasons in search of more returns with fewer injuries. MLS resembles a league-wide test bed. Orreco, selected for the 2026 Innovation Lab, combines AI, computer vision and biomarkers to optimise performance, predict injury and accelerate recovery. Technology proven inside the league may scale far beyond one club.

Collection itself is expanding. FIFPRO reports that tracking at the 2022 World Cup could capture up to 29 points on each player’s body, generating millions of data points in a single match. When position, speed, collision, injury and recovery data are connected, the result is not a scorecard but a continuously updated body file.

The better data predicts risk, the more easily it can predict value; the more it protects a career, the more it can shape a career opportunity.

02“For safety” is not a substitute for meaningful consent

A legitimate purpose does not automatically make a legitimate process. One-time consent is no longer enough. Athletes need to know what is collected, who can see it, how long it is stored, whether it trains a model and whether it can be deleted after exit.

The MLB collective bargaining agreement provides more specific boundaries. Wearable-device rules cover health and performance data on and off the field; use is voluntary in principle; clubs must disclose the technology and authorised personnel in writing; players can request copies and deletion; and commercial use requires consent.

The NBA G League takes a different position. Its 2025–26 agreement allows the league or a team to require approved devices during games, training and other activities, subject to negotiated rules.

The real institutional spectrum therefore runs from voluntary use to collectively negotiated mandatory use. The decisive questions are whether players shaped the rules, whether purposes are clear and whether refusal or withdrawal creates an indirect penalty. When a coach says monitoring will help an athlete get on the field, does a player competing for selection truly have room to refuse?

03A risk score that affects employment must be contestable

Injury prediction is not diagnosis or certainty. Training load, sleep, injury history, position and environment interact in complex ways; datasets can carry different biases across clubs, sports, genders and ages.

If a model labels an athlete high risk, a team may reduce training or reduce selection priority. One use protects the body; the other may affect bonuses, renewal and market value. Institutions — not software defaults — must draw the line.

FIFPRO groups player data into personal, performance, match technical, tracking, and health and biometric categories. Its research found that 80 per cent of professional players want access to their own data to improve performance, while also worrying about how it is collected and used.

Its proposed rights include information, access, withdrawal, restriction, portability, correction, complaint and deletion. Data collected through club equipment does not remove an athlete’s basic control over personal information.

IOC research on AI ethics in sport likewise emphasises human oversight, explanation, accountability and athlete welfare. At minimum, a model must not become a diagnosis by default; decisions affecting selection or contracts should state their basis; and athletes need a route to correct data and seek review.

Responsible sports AI does not only give an answer. It records who made the decision, on what basis, and how the athlete can challenge it.

04Commercial boundaries must be written before the partnership begins

Sports-technology partnerships usually begin with an easy objective: reduce injury, improve performance and extend careers. The harder questions come next. Can data move between clubs? Can it train a new model? What happens when the supplier changes? Can a commercial partner access derived insights?

FIFPRO argues that players need transparent information and participation in commercial uses of personal data. Participation should include fresh authorisation for a new purpose and a discussion of value sharing when the data produces an independent commercial asset.

Contracts between leagues, clubs and technology partners should answer at least five questions: who controls the data; who can access it; when a new purpose needs new consent; whether model outputs can enter selection and contract decisions; and what happens to raw data and derived models when the relationship ends.

This also changes the role of brands and agencies. Technology sponsorship cannot stop at delivering a device, screen or launch event. It must disclose data flows, permissions and exit mechanisms. If athlete data helps train a model sold across the industry, should athletes receive only the safety benefit, or also share in the commercial value?

The closer a technology partnership comes to an athlete’s body, the less its contract can be limited to features, price and exposure.

05The body remains the athlete’s

AI is making training more precise and placing the athlete’s body inside organisational decision-making with near-continuous frequency. Refusing all collection is not the answer; without data, injury prevention may remain dependent on intuition. But a safety purpose does not create unlimited data rights.

A more mature route writes governance into collective agreements, competition rules and technology contracts: collection has a purpose, access has limits, decisions can be explained, errors can be appealed, exit enables deletion and commercial use requires fresh authorisation.

The progress worth pursuing in sports AI is not seeing everything about an athlete. It is seeing more clearly while still recognising that the body belongs first to the athlete.