My guess is that this is part of some kind of machine learning pipeline, where users label edge cases to help train the model. Since it operates on account data, match data, and logs (see CSteamGPT_GetTask_Response), an anti-cheat use case would make sense, but it’s hard to say for sure.
This looks like data exchanged between the Steam client and server, and doesn’t contain any logic on its own.
It's not about "someone else probably will", it's about "someone else already has". No one is advocating for diffusion of responsibility.