What @lemon said is right. While $\bf F = -\nabla \phi\left({\bf r}\right)$ suggests that $\bf F$ depends on $\bf r$, it ultimately depends on time through ${\bf r} = {\bf r}\left(t\right)$. That is, ${\bf F} = -\nabla \phi\left({\bf r}\left(t\right)\right)$. So, there is no inconsistency in writing Newton's second law as (a dot represents a time derivative)
$$
{\bf F} = m \ddot{\bf r} = -\nabla \phi
$$
In fact, we don't have to restrict ourselves to a conservative force. $\bf F$ can depend on position, velocity, and time explicitly, ${\bf F} = {\bf F}\left({\bf r}\left(t\right),\dot{\bf r}\left(t\right),t\right)$, since you will end up with a perfectly consistent system of differential equations
$$
\begin{eqnarray}
\ddot{x} &=& F_x\left(x,y,z,\dot{x},\dot{y},\dot{z},t\right)/m \\
\ddot{y} &=& F_y\left(x,y,z,\dot{x},\dot{y},\dot{z},t\right)/m \\
\ddot{z} &=& F_z\left(x,y,z,\dot{x},\dot{y},\dot{z},t\right)/m \\
\end{eqnarray}
$$
To work a simple example, consider the potential energy
$$
\phi\left(x\right) = \frac{1}{2} k x^2.
$$
Then
$$
{\bf F} = -\nabla\left(\frac{1}{2} k x^2\right) = -k x \ \hat{\bf x} = m \left(\ddot{x} \ \hat{\bf x} + \ddot{y} \ \hat{\bf y} + \ddot{z} \ \hat{\bf z}\right),
$$
or
$$
\begin{eqnarray}
m \ddot{x} = -kx &\Rightarrow& x\left(t\right) = x\left(0\right) \cos\left(\omega t \right) + \frac{\dot{x}\left(0\right)}{\omega} \sin\left(\omega t\right), \omega = \sqrt{k/m} \\
m \ddot{y} = 0 &\Rightarrow& y\left(t\right) = {\dot y}\left(0\right) t + y\left(0\right)\\
m \ddot{z} = 0 &\Rightarrow& z\left(t\right) = {\dot z}\left(0\right) t + z\left(0\right)\\
\end{eqnarray}
$$