If virtual particles are the ones I know from Feynman diagrams, then momentum is still conserved at every vertex (interaction element). Unless the person who speculated that meant something else by "virtual particles," I don't see in the slightest how they could be used to obtain momentum or squirrel it away.
Besides, virtual particles have to go away at the end of every diagram... They can act as conduits between other particle fields but if a new particle remains after the interaction it will be just another reaction product (nothing virtual or spooky). Some go so far as to say that virtual particles don't particularly exist, but that's a philosophical statement I guess (they do indeed lack many properties of other things that exist). However I can't say much about that as the arguments for this-or-that nonintuitive ontology usually emerge from advanced theoretical research that is beyond my knowledge. (I.e. in momentum space virtual particles don't even remotely seem to exist, but maybe the picture seems more reasonable when you write it all down some other way.)
Virtual particles are particle-antiparticle pairs that are constantly being created and annihilated in empty space due to quantum ground-state variations. Supposedly their effects can be observed at the event horizons of black holes, giving rise to Hawking radiation, though this may be a simplification for the layman.
> though this may be a simplification for the layman
It's almost the opposite of that: a simplification for professionals' sake. You can set up an eternal black hole and enjoy the simplifications its time-invariance permits, but then you have to match the outgoing Hawking radiation flux (as seen at infinity) with an ingoing one.
With a collapsing-matter black hole you lose time-invariance, because, coarsely, there is a region of spacetime where there is spread out matter and no trapping surface ("event horizon", or whatever, let's be agnostic about that), there is a region of spacetime where there is a trapping surface, and finally there is a region of spacetime where the trapping surface is gone and there is a gas of thermal radiation. But on the bright side, then you don't need any ingoing flux to balance the outgoing Hawking radiation. Instead you look at non-trivial Bogoliubov transformations between these different regions, which model how observer in region-with-horizon sees particles that observer in region-with-uncollapsed-matter ("no horizon yet") does not. (You can additionally change the speed of collapse, which changes the difference in particles that the later observer sees: sufficiently slow collapse means no Hawking radiation at all.)
(Professionals even make other simplifications, like dealing with a free-as-in-linear scalar quantum field rather than realistic interacting matter; it won't matter much since "real" Hawking radiation would be almost only photons anyway, and the stark problems around the border between the region with the horizon (and Hawking radiation outside it) and the region with only Hawking radiation (and no horizon) likely do not depend on the precise content of the hottest Hawking radiation.)
Besides, virtual particles have to go away at the end of every diagram... They can act as conduits between other particle fields but if a new particle remains after the interaction it will be just another reaction product (nothing virtual or spooky). Some go so far as to say that virtual particles don't particularly exist, but that's a philosophical statement I guess (they do indeed lack many properties of other things that exist). However I can't say much about that as the arguments for this-or-that nonintuitive ontology usually emerge from advanced theoretical research that is beyond my knowledge. (I.e. in momentum space virtual particles don't even remotely seem to exist, but maybe the picture seems more reasonable when you write it all down some other way.)