All client-side anti-cheat systems are invasive, to varying degrees. Some restrict themselves to game files and data, others snoop on external processes and memory, and the worst (and most dangerous) even require administrator/root access or kernel drivers.

Yes, it’s terrible.

It’s also a divisive subject. Some gamers feel that cheating must be minimised at any cost. Others feel that their privacy and security are more important, and realise that only server-side measures can avoid circumvention anyway.

Unfortunately, there’s money being made selling client-side anti-cheat systems to game publishers, and implementing it in-house is cheaper/easier, so most will continue with that until enough of us reject it or we make it illegal.

implementing it is cheaper/easier than server-side

In many cases, 100% impossible. Take a first person shooter for example, how can you prevent wallhacking server-side? The only way is to not tell clients positions until the enemy is very close by, and then you get big problems with any kinds of lag interpolation etc.

In many cases, 100% impossible.

No, it is not.

If a particular kind of cheat is impractical to prevent in real time, it is sufficient to detect it and enforce consequences. (These can be banning, handicapping, isolating with other cheaters, or any number of other approaches.)

it is sufficient to detect it and enforce consequences

So how do you propose detecting that a client is rendering a model that shouldn’t be rendered (because it’s behind a solid wall for example) without some client-side anticheat?