The NBA plans to automate certain out-of-bounds decisions with cameras and AI, moving one of basketball’s objective calls away from on-court officials. Commissioner Adam Silver compared the proposed system with Hawk-Eye line technology in tennis.
The goal is an immediate possession decision without a replay stoppage. Referees would continue judging fouls and contact, where context and interpretation remain essential.
The distinction offers a useful model for sports automation: assign technology the narrow questions defined by physical position and preserve human authority over decisions that depend on intent, advantage and the flow of play.
A disputed playoff call shows the problem
Silver’s comments followed Game 5 of the Western Conference finals between the Oklahoma City Thunder and San Antonio Spurs. Victor Wembanyama was ruled to have touched the ball last, although replay showed it had bounced from Chet Holmgren’s foot.
Officials conferred but the ruling stood, and Oklahoma moved ahead 3–2 in the series. The incident illustrated how a visually objective event can still be missed at game speed and remain uncorrected under current review rules.
A Coach’s Challenge is the only route to review an out-of-bounds violation at any point in an NBA game. Each team begins with one and receives a second only after a successful challenge.
That system makes correction depend partly on whether a coach preserves a limited resource. Automatic detection could produce the same rule for every possession rather than linking accuracy to challenge strategy.
Optical tracking could make the decision instantaneous
The NBA began a multi-year partnership with Sony’s Hawk-Eye Innovations in 2023 after testing 3D optical tracking at Summer League and league arenas. Cameras positioned around the court can reconstruct the ball and player locations from several angles.
An automated system would determine which player last touched the ball before it crossed the boundary and immediately assign possession. Silver’s ambition is simple: no conference, no replay delay and no interruption to the game.
Other sports have already separated objective geometry from judgement. Tennis uses electronic line calling, FIFA employs semi-automated offside technology and Major League Baseball is introducing an automated balls-and-strikes challenge system in 2026.
Basketball presents its own difficulty because hands, feet and the ball can converge rapidly near a line. The system must handle blocked camera views, deflections and extremely small timing differences without slowing play.
The NBA already operates a Replay Centre in Secaucus, New Jersey, connecting all 30 arenas to 94 high-definition monitors and 23 workstations. Automated possession decisions could use that infrastructure while reducing the number of routine cases requiring manual attention.
Human referees retain the judgement calls
Silver said officials would remain responsible for fouls. Contact occurs on many plays, but a referee must decide whether it affected movement or prevented a player from continuing.
That boundary matters for trust. Technology performs best when the rule can be translated into observable position and timing. It is less suited to interpreting severity, context or whether incidental contact changed the play.
The league will need transparent evaluation before rollout. Tests should measure accuracy across arenas, camera failures and unusual deflections, while procedures define what happens when the system is uncertain or unavailable.
Players, coaches and viewers also need a clear signal that explains the result without introducing another lengthy review. An automatic call that cannot be trusted may simply move disputes from referees to the technology.
Silver offered no fixed launch date but said the NBA expects to move fairly quickly. A staged introduction in limited games would allow the league to compare automated decisions with expert review and refine failure handling.
The proposal is not an attempt to remove referees from basketball. It is a division of labour: machines handle a narrow spatial fact, while people judge the human contest around it.
If the tracking proves fast and dependable, out-of-bounds automation could improve accuracy while preserving the game’s rhythm. Its success will depend on staying within that carefully defined boundary rather than treating every disputed call as a problem AI should solve.
