Got some folks talking past each other.
First of all, recoil begins as soon as the bullet begins to move. It has to, that's what Newton's law is all about. Unless the rifle is clamped in such a way that it can not move at all then it must move as soon as the bullet starts moving.
This can be readily verified via slow-motion videos--which is a good thing since that's what science (physics) tells us must happen.
But does that motion affect the bullet's trajectory? MAYBE.
If the recoil of the rifle is straight back--the muzzle doesn't rise or drop or move to the left or right then there will be no effect on the trajectory.
If the recoil motion ends up causing the rifle's muzzle to rise, drop or move left or right then there will be an effect on the trajectory.
Let's say that the muzzle does move during recoil. Does that mean the rifle won't be accurate? Actually the rifle can be very accurate even with the recoil motion causing muzzle rise. The recoil motion happens very fast and the forces involved are high enough that it's difficult to effectively resist them in the very short amount of time that the bullet takes to exit the barrel. What that means is that the motion due to recoil will usually be very consistent. Even though the recoil affects the trajectory,
it affects it the same way every time. So the bullet hits in the same place on the target every time and the rifle is very accurate in spite of the motion of the muzzle during recoil before the bullet exits.
In revolvers, the amount of muzzle rise due to recoil is so significant that it's often easy to note that the line of the sights is aiming significantly higher than the barrel is. That is, the barrel is pointed noticeably downward when the sights are aligned on the level. The motion of muzzle rise drives the barrel upwards as the bullet moves down the barrel, and by the time it exits, the bore is pointing where the sights were pointing. Because the barrel is fixed to the frame and the frame is restrained by the shooter's hand below the bore, as soon as the recoil motion starts, the muzzle also starts to rise.
This effect isn't seen as much in locked breech autopistols with floating barrels because during the time the bullet is in the barrel, the barrel (and slide) are recoiling nearly straight backwards. Muzzle rise doesn't start until after the bullet has left the barrel. The slide/barrel start moving straight back immediately but the muzzle doesn't rise until after the bullet's gone.
Here's a diagram I made that shows the relationships between sight line and bore line for a revolver and several autopistols. Note that the revolver bore is pointed downward compared to the sights while the autopistol bores align well with the sights.
But revolvers are sort of an extreme case. The gun is relatively light and is restrained only by the shooters hand which is well below the bore. That results in a lot of muzzle rise very early in the recoil cycle.
Rifles tend to recoil straight back without a lot of muzzle rise because the stock transfers the recoil to the shooter's shoulder which is pretty close to being right behind the muzzle. What little muzzle rise there is tends to be pretty consistent given the way rifles are held. So it's not much of an issue except in extreme cases.
So yes, recoil motion starts with the bullet still in the bore. Has to. But that doesn't imply inaccuracy. It's not even a given that the recoil motion will affect the trajectory-- if the rifle recoils straight back with no muzzle rise then there's no effect. If there is muzzle rise, it tends to be consistent from shot to shot which means that once the rifle is zeroed the consistent effect is taken into account and is a non-issue.