This is an open problem. No example is known; nor is a proof that there is none.
Open Problem Garden says:
> This question was first asked in 1984 by Martin LaBar and popularized in 1996 by Martin Gardner, who offered \$100 to the first person to construct such a square. In 2005 Christian Boyer offered €1,000 and a bottle of champagne for a solution to a somewhat easier problem. For a review of the history of research, see […]. For basic facts about the anticipated $ 3\times 3 $ magic square of squares, see […].
Some discussion is given on mathpages.com.