@Michaelp800 Could this solution also not be abused by say a group of people who have an issue with a particular person?
For example, say Player ABC went 102 - 0 in a tank in the previous match and during that match they kept wrecking a platoon with say 8 - 10 people on the other side. As you're relying on an AI system that relies on player input, how would the system know what is true and what is false if all 8 - 10 people spammed reports of racism, hate speech, etc? Can you see where this would become a very slippery slope.
While in theory your idea works, it heavily relies on players being honest and not abusing a quick 'boot' option to high K:D players on the other side they don't like. Not to sound cynical but I wouldn't trust 99.99% of the playerbase of any game to be even 10% honest lol.
For any form of honesty you'd need to have actual admins monitoring the games as relying on an AI system will come down to the amount of reports made if keywords aren't typed. How does the AI system know if what they said was verbal or not, it can't and so this is where the idea fails unfortunately