I think it's worth expanding a bit on Hal's answer to try and make it a bit less technical. We denote a point in spacetime as $(t, x, y, z)$ i.e. both the position $x, y, z$ and the time $t$. In the absence of time machines we can only pass through a spacetime point once. Of course you can go back to the point in space $x, y, z$ but only at a later time so you can't get back to the point $(t, x, y, z)$. If it were possible to pass through $(t, x, y, z)$ go somewhere else then get back to $(t, x, y, z)$ your trajectory would form a loop, and we call this a closed timelike curve (or CTC). It's closed because it's a loop and timelike is a technical term that means you don't have to travel faster than light to go round the loop.
For any particular CTC there will be some earliest time that lies on the loop, so by going round the loop you can only get as far back in time as this earliest point. The point that Morgan Freeman is making is that for all the types of time machine we know about this earliest point corresponds to the creation of the time machine. So the statement is true for all the time machines we know about. I don't know if there is a general rule that says it must be true for all time machines, but I suspect not.