Solution:
We proceed by induction on N.
Base case: N=1. The single black square can be cut out as a 1×1 square. The black area is 1, which is between 1/5 and 4/5 of 1 (since 1=1). So the base case holds.
Inductive step: Suppose the statement is true for all numbers less than N. Consider N black squares.
Let k be the largest integer such that k2≤N. Consider all possible k×k squares on the grid. Each such square contains at most k2 black squares. Among all k×k squares, choose one that contains the largest number m of black squares. Then m≥N/(n−k+1)2, where n is the side length of the grid (but since the paper is large, we can always find such a square).
If m≥k2/5, then m≤k2 and m≥k2/5. If m≤4k2/5, then we can cut out this k×k square, and the black area is between 1/5 and 4/5 of its total area. Remove these m black squares, and apply the induction hypothesis to the remaining N−m black squares.
If m>4k2/5, then m is close to k2. Cut out this k×k square. The black area is more than 4/5 of its total area, but less than or equal to k2. In this case, we can cut out a larger square. Consider a (k+1)×(k+1) square containing the k×k square. The number of black squares in this (k+1)×(k+1) square is at most N, and at least m. If the number of black squares in this (k+1)×(k+1) square is between 1/5 and 4/5 of its area, we are done. If not, repeat the process with larger squares.
Since the total number of black squares is N, and the area of the paper is much larger, we can always find such squares. At each step, we remove a square with the required property, and the process terminates after finitely many steps.
Therefore, it is always possible to cut out a set of disjoint square pieces of paper so that all the black squares are removed and the black area of each piece is between 1/5 and 4/5 of its total area.