This has an elegant proof using universal nets (or ultrafilters, or convergence of maps with respect to ultrafilters which would be my favorite way to do it). To show the inverse map is continuous, it suffices to show it preserves convergence of universal nets. So suppose $(g_i)$ is a universal net converging to some $g\in G$. The net $(g_i^{-1})$ is then also a universal net, and so since $G$ is compact it must converge to some $h\in G$. Now the net $(g_i,g_i^{-1})$ converges to $(g,h)$ in $G\times G$, so by continuity of $m$, $(g_ig_i^{-1})$ must converge to $gh$. But $g_ig_i^{-1}=1$ for all $i$, and so since $G$ is Hausdorff this constant net can only converge to $1$. Thus $gh=1$, and $h=g^{-1}$. That is, $(g_i^{-1})$ converges to $g^{-1}$, as desired.
More generally, a similar argument shows that if $Y$ is compact and $f:X\to Y$ is a map whose graph is closed in $X\times Y$, then $f$ is continuous. In this case, the graph of the inverse map is closed in $G\times G$ since it is $m^{-1}(\{1\})$.