ATP's AI hides abusive comments so its safety team can chase the real threats
162,0003.1M+ social media comments scanned in the first year; 162,000+ flagged as severe, aimed at 245 players; abusive comments hidden as they are detected; 3,300+ comments escalated, 68 perpetrators identified and 28 law enforcement referrals (ATP figures).
“In its debut year, the ATP’s Safe Sport initiative has scanned more than 3.1 million social media comments, flagging over 162,000 targeted at 245 players as “severe” and hiding abusive comments as they are detected, in a bid to shield the leading ATP players from harmful digital content.”
ATP press release, 21 Aug 2025 (also BBC Sport, 21 Aug 2025), 21 Aug 2025
- Who
- The ATP, which runs the men's professional tour (ATP Tour and Challenger Tour). Safe Sport covers its Group 1 players (top 250 in singles, top 50 in doubles) and the ATP's own social channels. Built with Tennis Data Innovations (TDI), Sportradar and Arwen AI.
- Problem
- More than 1 in 10 comments on players' posts were abusive, rising to 50% on some players' pages, according to the ATP. That's millions of comments a year: no team can read them all by hand, and the players were the ones seeing them.
- Agent
- Safe Sport uses real-time AI to detect and hide abusive comments across players' social media accounts as they are posted. The serious cases go to Sportradar's Integrity Services team, who investigate, identify the senders and refer threats to law enforcement. The programme now also helps players facing email threats, impersonation, deepfakes and scams.
We'll build it for you
Book a callView the prompt
Try it yourself with Arwen AI
You run social media for a tennis or padel club, academy or tournament. Goal: set simple rules for an AI tool that screens comments on our accounts, hides the abusive ones and passes the serious ones to a person. My input: [the last 50 comments on our posts and our players' or coaches' posts, copied as-is], [our community rules], [the accounts to cover], [who on our team handles threats and how to reach them]. Give me: 1) the comments sorted into fair criticism, spam, abuse and threats; 2) a hide, keep or escalate rule per group, with examples; 3) a short message template for the person on duty when something is serious, and when to involve the federation or the police. Never treat criticism of results as abuse. When in doubt, send it to a person.What we can do for your club or brand
Clubs, academies and tournaments get abusive comments too, often under posts about juniors or coaches. The Court can set up an AI filter on your club's or tournament's social accounts that hides abusive comments as they come in and sends the serious ones to a named person on your team, so your community manager stops reading it all and your players don't have to.
Source:
ATP press release, 21 Aug 2025 (also BBC Sport, 21 Aug 2025), 21 Aug 2025 · first-party source
What they did
In July 2024 the ATP launched Safe Sport with Tennis Data Innovations, Sportradar and Arwen AI for its top players and its own channels. The AI scans comments and hides abuse in real time; Sportradar's Integrity Services team takes the serious cases further, up to police referrals. The ATP has since extended it to email threats, impersonation, deepfakes and scams, and added digital-safety lessons through ATP University.
How it went
First-year figures (ATP, Aug 2025): 3.1M+ comments scanned, 162,000+ flagged as severe across 245 players, 3,300+ escalated, 68 perpetrators identified and 28 law enforcement referrals. The AI takes the volume, and people spend their time on the threats that need action. Player council member Dušan Lajović says his social media is now "defined by who I am, not by my performance on the court". Limits: figures are self-reported by the ATP; BBC Sport notes the system did not catch every instance of abuse, and the ATP says it won't eliminate abuse overnight; it only covers the top 250 in singles and top 50 in doubles, so players further down aren't covered by this service; no second-year figures published yet.