AI-powered highlights are changing the game

For the Paris Olympics, Peacock said it could stream nearly 7 million personalized versions of its daily recap, each narrated by an AI re-creation of Al Michaels’ voice. Two years later, AI highlights and AI-written commentary have spread from that one experiment into league apps, tennis apps and live broadcasts. MLB adds Gemini-written notes to its pitch-by-pitch feed. The NBA uses AI to pull comparable plays from thousands of games. ESPN publishes AI-written recaps under its own byline. Each step brings fans more tailored coverage and gives editors more output to check.

How automated AI highlights get clipped

The plumbing comes first. Vendors such as WSC Sports scan a live game feed, find key plays and cut clips for social and digital channels as the game unfolds. When the NWSL signed with WSC in May 2024, the company said it would analyze more than 150 league games that season and produce “thousands of hours” of content, Sportcal reported. The same report listed the WNBA, the LPGA Tour and Spain’s Movistar Plus among WSC’s clients.

Automated clipping answers a volume problem: the NWSL deal alone covered more than 150 games in a single season. WSC’s service builds those videos in real time and pushes them to the league’s social and digital channels.

Personalized highlights at Olympic scale

Peacock’s “Your Daily Olympic Recap” showed what personalization can look like. According to NBC Sports’ announcement, subscribers picked up to three favorite sports plus topics such as viral moments or behind-the-scenes stories. Each day the system drew on hundreds of clips to build a playlist of about 10 minutes. Generative AI wrote the narration and voice synthesis delivered it in Michaels’ voice, trained on his past NBC broadcasts. NBCU editors reviewed the audio and clips before release.

Michaels said he was “skeptical but obviously curious” before he heard a demo. NBC said the 5,000 hours of live coverage from Paris could support nearly 7 million personalized versions in the U.S.

The NBA has built personalization into its data layer. Through its October 2025 partnership with AWS, the league launched Play Finder, which uses AI to analyze player movement across thousands of games and retrieve similar plays on demand. The league plans to bring future versions to the NBA App. You can read how that tracking data works in our feature on the next era of player tracking.

AI commentary experiments from MLB to Wimbledon

AI now writes the words around the clips too. MLB launched Scout Insights for Opening Day 2026, adding Gemini-generated notes to the Gameday pitch-by-pitch feed in its app and on MLB.com, Sports Video Group reported. The insights cover pitcher-batter matchups, physical feats and odd game situations.

Speed shaped the design. Engineers found that generating each note live took too long, so the system predicts likely game events from the day’s lineups, writes insights in advance and matches the best one to what happens on the field. Google Cloud built a “surprisal” score to rank which facts are interesting enough to show. One beta example from Google’s announcement noted that Jordan Walker hit a 114.3 mph single, the ninth-hardest single at American Family Field since 2014.

Tennis took a conversational approach. For Wimbledon 2025, IBM and the All England Club launched Match Chat, an assistant inside the Wimbledon app and website that answers fan questions during live singles matches, per IBM’s announcement. The tool runs on watsonx Orchestrate and IBM Granite models trained on Wimbledon’s editorial style. An updated Likelihood to Win feature shows a projected win percentage that shifts through each game.

Broadcasters use AI for on-air context as well. Prime Video’s Prime Insights adds real-time NBA stats and flags mismatches during games, and the same feature runs on its NFL, NASCAR, Champions League and Masters coverage, Sports Video Group reported. The NBA’s Play Finder also sends real-time alerts to help commentators add historical context.

Where AI recaps go wrong

ESPN offers the clearest record of the risks. The network began publishing AI-written recaps in September 2024 to cover under-served sports, including the PLL, NWSL, NCAA baseball and softball, the G League and NBA Summer League, Front Office Sports reported. The stories carry the byline “ESPN Generative AI Services.”

The program has stumbled in public. A 2024 recap of Alex Morgan’s final pro match left out Morgan and her farewell. In July 2026, a Summer League recap headline suggested the Nuggets had beaten the Rockets, when Houston won 97-86. The first fix introduced a new error before ESPN corrected it again. An ESPN spokesperson blamed human editing and said no article goes live until an editor reads it and publishes it by hand. During the NBA Finals, ESPN also dropped AI-generated moving portraits after an image of Tony Parker drew backlash.

Both NBC and ESPN built human review into their AI programs. NBCU editors checked the Peacock recaps, and ESPN editors publish each AI story. In ESPN’s July case, the company said the errors came from that human step.

What AI highlights mean for fans and creators

For fans, AI highlights mean clips and recaps built around the teams and players you pick. ESPN pitched its AI recaps as a way to cover under-served sports such as the NWSL and the G League, and WSC’s NWSL deal put real-time clips of every match on the league’s own channels.

Human voices still drive how fans talk about a game. On FansPulse, creators call games live and fans react in real-time chat rooms, a format that pairs well with machine-made clips. For the business side of these tools, see our coverage of the next wave of sports streaming and the wider AI section.