Riot Games will begin monitoring players voice communications on July 13th, according to PC Gamer. The language models that the game company will use to evaluate player reports will be trained.
Riot's privacy policy was updated in April 2021. The new terms allow Riot to record and possibly evaluate voice data in order to combat hate speech and harassment over voice chat. Riot will analyze recordings when a player complains about someone. This should help the company determine if the player violated its policies.
The only way to opt out of this system is to disable voice chat completely
Riot isn't going to start assessing player reports based on these recordings just yet, it's using the information it gathers to help build the prototype of the system it expects to roll out later this year Riot will only look at the conversations of English-speaking players. If you don't want to use voice chat, you can use another communication tool.
Riot notes in its announcement that they have systems in place to make sure they can correct any false positives if there is a mistake.
Riot won't actively monitor your live game comms and will only listen to and review voice logs if you're reported for disruptive behavior This information will be deleted after it resolves the situation, like it does for reports made over its chat systems. Privacy is bound to be a concern for some players, like the always-on Vanguard anti-cheat system that monitors your activity both in and outside of Valorant.
The reporting system is one of the ways that Valorant is trying to crack down on toxic players. Riot started allowing players to add specific words to a "muted words list" that is supposed to help block out abusive content in chat.