Microsoft has long wanted to get vendors out of the kernel. It’s a huge privacy/security/stability risk, and causes major issues like the Crowdstrike outage.
Most of those issues also apply to kernel anti-cheat as well, and it’s likely that Microsoft will also attempt to move anti-cheat vendors out of kernel space. The biggest gaming issues with steamOS/Linux are kernel anti-cheat not working, so this could be huge for having full compatibility of multiplayer games on Linux.
Basic anti-cheat already does this, but also with memory, because most cheats are reading/modifying what is in memory. I think the only ethical solution for anti-cheat is on the server side, with machine learning perhaps, kind of like VACnet.
Yeah and a lot of cheats know the anti cheat is checking memory so they also modify the anti cheat and essentially mess up their memory check to fool it into thinking nothing has been modified. It’s just a cat and mouse game where the cheats bypass the anti cheat and the anti cheat adding more detectors.
The problem is that, with a good enough cheat, it can be impossible to distinguish from a very good player.
The best cheats use a secondary device emulating human input and reactions, which is practically undetectable.
You will never stop cheaters, ever. It’s something we have to live with. It’s annoying when it happens, but it’s hardly the end of the world either.
So I’d rather have the AC running on the server and not invading my system.
A secondary device can’t be identified by kernel level anti-cheat either. If you have a standalone device that identifies as a USB keyboard and mouse and then generates inputs that give you a 100% headshot count, there’s nothing you could detect through the kernel, since all it detects are keystrokes and clicks.