**Hint:** You can treat $A$ and $B$ as (dependent) random variables. Then it is asked to calculate
$$\sigma_{AM}=Cov\left(A,\frac12(A+B)\right)$$
where $Cov(A,A)=\sigma_A^2$ and $Cov(A,B)=\sigma_{AB}$
$\texttt{Additional hint}:$ You can derive the result straightforward if you use the property that the covariance is $\color{blue}{\texttt{bilinear}}$.