Any point can be used for the calculation of the moments, however there are more considerations compared to the static case. This is easier with an example:
Assume the case of cylindrical mass which has a wire around it and its left to drop. In that case

The left is the Free body diagram, and right is the kinetic diagram (which has the accelerations).
If you write the equations of equilibrium for x, and y (just for completeness) you get:
$$ x:\qquad 0 = m \cdot a_x \tag{eq.1} $$
$$ y:\qquad T- W = m \cdot a_y \tag{eq.2}$$
This does not change for any point that is selected for the moments. However for the moments you can do one of the following things (there are probably more):
- select the center of mass, and use the mass moment of inertia about the center of mass G (this is the simplest).
- select another point, and use the mass moment of inertia about the center of mass G. in that case you need to consider the inertial forces.
- select another point (let's call it P), and use the mass moment of inertia about that point P. in that case you the inertial forces are consider around that point.
All of the above are (almost miraculously) equivalent.
1. moment about G
$$\sum M_G = I_G \cdot \alpha\Rightarrow -T\cdot r = I_G \cdot \alpha$$
where:
- $I_G$ is the mass moment of inertia about G (and its equal to $\frac{1}{2}mr^2$
- $\alpha$ is angular acceleration
$$-T\cdot r = \frac{1}{2} mr^2 \cdot \alpha \tag{eq.3}$$
2. moment about P using $I_G$ mass moment about G
Assume the following point P

In that case:
$$- W\cdot r = I_G\cdot \alpha +\color{red}{m\cdot a_y} \cdot r \\Rightarrow $$
$$- W\cdot r - \color{red}{m\cdot a_y} \cdot r = I_G\cdot $$
$$ - (W + m\cdot a_y) \cdot r = I_G\cdot \alpha $$
However from eq.2: $ T = m \cdot a_y + W$. So we end up with
$$ - T \cdot r = I_G\cdot \alpha \tag{eq.4}$$
Which is the same as eq.3.
3. moment about P using $I_P$ (mass moment about P)
if the take the point P and the mass moment about P, then the inertial forces are drawn around P. In that case:
$$- W\cdot r = I_P\cdot \alpha \\Rightarrow $$
(i.e. we don't need to consider the inertial forces like in case 2). However in that case:
- $I_P$ is equal to $I_P = \frac{1}{2}m r^2 + m*r^2$ (parallel axis theorem).
So:
$$- W\cdot r = (\color{red}{\frac{1}{2}m r^2} + m\cdot r^2)\cdot \alpha \\Rightarrow $$
$$- W\cdot r = (\color{red}{I_G} + m*r^2)\cdot \alpha \\Rightarrow $$
$$- W\cdot r = \color{red}{I_G}\cdot \alpha + \color{green}{m\cdot r^2\cdot \alpha}
\tag{eq:5.a}$$
With a little rearranging the term $\color{green}{m\cdot r^2\cdot \alpha}$, becomes:
$\color{green}{m\cdot r^2\cdot \alpha} =(m\cdot r\cdot \alpha)\cdot r = m\cdot (r\cdot \alpha)\cdot r \tag{eq:6a}$
However because there are also kinematic constraints. e.g. in this particular problem the relationship between $\alpha$ and $a_y$ is
$$\alpha \cdot r = a_y$$
therefore, eq.6a becomes: $$m\cdot a_y \cdot r \tag{eq:6b}$$
By substituting eq. 6b in eq.5a we obtain:
$$- W\cdot r = \color{red}{I_G}\cdot \alpha + m\cdot a_y \cdot r
$$
$$- W\cdot r - m\cdot a_y \cdot r =I_G\cdot \alpha +
$$
$$- T\cdot r = I_G\cdot \alpha \tag{eq:5.b}$$
which is equivalent to equation eq.3.