On the failure of Von Neumann's inequality