This is roughly what Valve does for CS2. But, as far as I understand, it's not very effective and unfortunately still results in higher cheating rates than e.g. Valorant.
The example still kind of applies. In the CS world, serious players use Faceit for matchmaking, which requires you to install a kernel-level anticheat. This is basically what you're suggesting, but operated by a 3rd party.
But so far that still seems to be miles away.
The primary one is a standard user-mode software module, that does traditional scanning.
The AI mechanism you're referring to is these days referred to as "VAC Live" (previously, VACNet). The primary game it is deployed on is Counter-Strike 2. From what we understand, it is a very game-dependent stack, so it is not universally deploy-able.
But anyway counterstrike did have community policing of lobbies called overwatch - https://counterstrike.fandom.com/wiki/Overwatch
It was terrible as it required the community to conclude beyond reasonable doubt the suspect was cheating, and cheats today are sophisticated enough to make that conclusion very difficult to make
And the way community policing worked in the past is that the "police" (refs) could just kick or ban you. They don't need a trial system if the community doesn't want that.
I guess I didn't exactly make that clear...
A few of the arguments advanced by the "anti-anticheat" crowd that inevitably pops up in these threads are "anticheat is ineffective so there's no point to using it" and "anticheat is immoral because players aren't given a choice to use it or not and most of them would choose to not use it".
I don't believe that either of these are true (and given the choice I would almost never pick the no-anticheat queue), but there's not a lot of good high-quality data to back that up. Hence, the proposal for a dual-queue system to try to gather that data.
Putting in the community review of the no-anticheat pool is just to head off the inevitable goalpost-moving of "well of course no system would be worse than a crappy system (anticheat), you need to compare the best available alternative (community moderation)".
My understanding of the proposal is that it advertises no invasive anticheat (meaning mostly rootkit/kernel anticheat). So, the value proposition is anyone who doesn't want a rootkit on their computer. This could be due to anything from security concerns to desiring (more) meaningful ownership of one's devices.
Community moderation simply doesn't work at scale for anticheat - in level of effort required, root cause detection, and accuracy/reliability.
I rather play with cheaters here and there than install some kernel level malware on machine just to make sure EA, Activision, et al can keep raking in money hand over fist.
Or better yet, I can just play on console where there is no cheating that I have ever seen.