Sure, systems like this have been used at least 15 years by online poker bots. Poker software has some of the strongest anti-cheat systems in the world. The arms race didn't stop there, I know PokerStars moved on to behavior modeling about a decade ago. Now with machine learning being mainstream, mainstream games are starting to do behavior modeling as well. [1][2]

However most games either don't have any anticheats at all, or have very limited ones. Specifically you can run most games under a normal user and run the cheat under an administrator user. Let Windows help hide it.

--

[1] Valve has used deep learning for CSGO https://www.pcgamer.com/vacnet-csgo/

[2] Third parties have done work for Dota 2, with hints of official Valve variants coming as well https://www.reddit.com/r/DotA2/comments/8816oh/12_of_all_mat...

I've heard rumors of such prepackaged systems being sold on the high-end (for people willing to spend ~$1000 just to cheat in a video game...). But VAC and similar anticheat aren't even at the level of being able to worry about such attacks yet; they're still desperately trying to solve the 99% case of free-to-download cheat software on the first page of google search results.

That's more like "bots", rather than "cheats", but yes.

I remember something similar for Eve Online bots many years ago. That solution was capturing video frames straight from the video card(so, detectable), but that's basically the same idea.

Monitoring in game behavior can easily detect naive bots, which are most of them(you are logged in 24/7 doing the same thing).

No but it used to be pretty common to intercept/modify UDP packets bi-directionally and build simple aim bots/etc.

Sure, this approach is used, but the main problems are latency and relative development cost.