Yes. And this effect is measurable and map-able, because it results in E-mode polarization of the CMB being converted into B-mode on scales (a few arcminutes to many arcminutes) where this is the dominant source of B-mode polarization. B-mode polarization on arcminute scales is therefore a direct measure of lensing of the CMB. This is a subject of current research, and there are now some regions of the sky (a few hundred square degrees) where it has been mapped.
Current CMB anisotropy measurements are just deep enough so that the lensing effect of a single cluster of galaxies can be seen in the CMB, provided the CMB background in that direction happens to have a gradient. The lensing effect is seen as a "dimple" in the gradient at the position of the galaxy cluster. This measure of gravitational lensing by the cluster can then be compared to the lensing effect on background galaxies observed at visual wavelengths. These measurements are noisy, but within that noise no inconsistencies have been found.
One reason this effect is being intensively studied is that, in order to measure "r" --- the "tensor-scalar ratio of the Early Universe", it is essential to correct for the amount of B-mode polarization that results from gravitational lensing and subtract it out from the degree-scale measuremnts of B-mode, a scale where primordial gravitational waves may dominate.