First have a look at my answer to Black holes and positive/negative-energy particles for some background on Hawking radiation. The pairs of virtual particles analogy is just an analogy and not what actually happens. In fact virtual particles don't really exist in the way that real particles do - see this article by Matt Strassler for more on this.
But your question does touch upon a mystery in quantum field theory so it's worth a bit of discussion. In QFT the vacuum has a zero point energy, however since in QFT we only ever measure energy differences this zero point energy has no immediate physical effect. However like everything in quantum mechanics the zero point energy is subject to the uncertainty principle so repeated measurements of the vacuum energy will return different answers. These fluctuations in the measured vacuum energy are what is normally modelled by virtual particles. Let me emphasise once again that these virtual particles are a computational device. The energy fluctuations are real, and lead to phenomena like the Casimir effect, but they aren't really due to particles jumping in and out of existence.
The mystery comes when we move to general relativity because although in QFT we are only interested in energy differences, in GR it is the absolute value of the energy density that curves spacetime. The problem is that if we attempt to calculate the gravitational effect of the vacuum energy the answer we get is 120 orders of magnitude too big.
Currently there is no good explanation for this discrepency and we have to admit we don't understand what is going on here. The pragmatic approach, favoured by many (most?) of us, is to assume that for some reason that we'll discover soon vacuum energy doesn't gravitate and the problem goes away.
Finally we should note that even if vacuum energy did gravitate it would not behave like dark matter but instead it would behave like dark energy and cause an exponential expansion of the universe.