Your approach is essentially correct. The linear system of equations
$$\left\{ \begin{eqnarray}
a_1 &+& a_2 &+& \dots &+& a_n &=& 0\\
\alpha_1 a_1 &+& \alpha_2 a_2 &+& \dots &+& \alpha_n a_n &=& 0\\
\alpha_1 ^2 a_1 &+& \alpha_2 ^2 a_2 &+& \dots &+& \alpha_n ^2 a_n &=& 0\\
\vdots\\
\alpha_1 ^{n-1} a_1 &+& \alpha_2 ^{n-1} a_2 &+& \dots &+& \alpha_n ^{n-1} a_n &=& 0\\
\end{eqnarray} \right.$$
having the unknowns $a_1, \dots, a_n$ has for determinant the Vandermonde determinant
$$\begin{vmatrix}
1 & 1 & \dots & 1 \\
\alpha_1 & \alpha_2 & \dots & \alpha_n \\
\alpha_1 ^2 & \alpha_2 ^2 & \dots & \alpha_n ^2 \\
\vdots\\
\alpha_1 ^{n-1} & \alpha_2 ^{n-1} & \dots & \alpha_n ^{n-1} \\
\end{vmatrix} = \prod _{1 \le i, j \le n} (\alpha_i - \alpha_j) .$$
If $\alpha_i \ne \alpha_j$ for $i \ne j$, then the determinant is non-zero, hence the solution $a_i = 0$ for all $1 \le i \le n$ is the only one.
If there exist $i$ and $j$ with $\alpha_i = \alpha_j$, then the determinant is $0$, hence there exist non-trivial solutions.
Notice that allowing $m$ to be arbitrary in $\Bbb Z$ was way too much, since we only used $m \in \{0, \dots, n-1\}$.