From @PearsonArtPhoto's answer herehere, he estimates a beamwidth of about $3.5~\text{$\mu$rad}$, or roughly $0.002~\text{deg}$ (unfortunately, his main source is also offline because of the government shutdown). He goes on to point out that this corresponds to an area on the Earth of about $4.9~\text{km}^2$.
While this does indeed levy a tighter pointing requirement, it's still about 10 times larger than typical absolute pointing accuracy seen in ground-based telescopes (about 1 arcsec).