Yes, it is totally possible. You need to use classical GR with continuous matter distributions and it requires energetic continuous matter with zero rest mass, and it is an unstable setup, so you have to do it 100% perfectly with absolutely no margin for any error. Let's see how.
You can take a sphere of radius $R$ in Minkowski space, and take a Schwarzschild solution where the event horizon is a spherical shell of surface area $4\pi R^2$. Cut out the interior of the Schwarzschild solution and glue the exterior and the sphere together (that's possible since they have the same surface area). The discontinuity implies there is a source term at that surface, so you need to place energetic matter there, and it has to fit on the shell perfectly. It could be almost any matter on the shell if you only want the black hole to be empty just for a little bit, but in that case it's only a shell that hasn't yet formed the singularity.
If you want the black hole to stay empty, then since the matter must stay on that shell it has to be moving outwards radially and move at the speed of light, so it must be energetic but have zero rest mass. Then it just sits on the event horizon forever. It tries to move away at the speed of light, but the event horizon is the exact place where when you do that you stay on the same surface of surface area $4\pi R^2$.
This situation is unstable because any slight perturbation will cause it to collapse into a singularity. It is a black hole because there is a mathematical event horizon, but it is a completely empty (and flat) spacetime on the inside. It is only a mathematical event horizon since if anything actually tried to cross the event horizon that would trigger the collapse into a singularity, so if you want an event horizon with emptiness on the inside, you can't let anything cross it (well, that's fair because then it wouldn't be empty on the inside).
My explanation is done in the style of just writing down a manifold, then computing what the Einstein tensor is, then finding a stress-energy tensor that can equal it, the manifold-first style. In this case the stress-energy tensor doesn't violate any energy conditions, is singular in the sense that a finite amount of energy is confined to a vanishingly thin spherical shell, and has no rest mass since the source moves outwards at the speed of light. Known gravitational sources without rest mass (just energy, momentum, stress etc.), such as classical electromagnetic waves, can't be confined to a vanishingly thin shell, because they need space in which to undulate in order to propagate. So I see no known way to build this shell (and since it is unstable, it couldn't be done in practice anyway).