Oh yeah, and for behaviour detection of this, it’s kinda annoying they don’t detect it because I don’t think it would be difficult to do this (either from a problem solving perspective or the amount of computational power that would be required).
Just track the x and y deltas and their derivative over time (in this case, the derivative is just the difference between the current sample and the previous one, so no calculus required, just a subtraction per sample). Then check if they are continuous. X and y deltas are velocity, which must be continuous because the mouse is a physical object and subject to inertia. Acceleration should also be continuous because of the limitations of our muscles (though if your mouse bumps your keyboard or your hand is moving and bumps your mouse, you can see natural acceleration that isn’t continuous, but these wouldn’t directly preceed a successful shot at a target).
Then just watch for spikes in either of those. A better cheat program could smooth the spikes, but it slows down the capability of the aim bot.