It increases.
I should clarify some things before I get to the actual "why."
There are 2 opposing tendencies of spanwise flow. (well I'm sure there might be more, but these are the 2 basic ones) The first one I'm sure you're already familiar with: Based on air's tendency to flow from high to low pressure, it flows inward on top of the wing and outward below the wing, since maximum lift (and pressure differential) is in the middle, and there's a lift dropoff near the tips. That's the prevalent one on straight wings.
Now, if you sweep the wing, what happens? Air still wants to go to low pressure. Let's follow the path of a parcel of air meeting the wing near the forward end, on the top surface. We know from before that pressure is ambient in front of the wing, lowers right above the wing, and goes back up to ambient somewhere after the wing. Well if we're that parcel of air right above the wing, straight ahead there's a pressure increase back up to ambient. But if we turn toward the wing, there is lower pressure. So we turn that way, to go where the low pressure is. That is why on swept wings there is crossflow outwards on top of the wing.
Sorry that explanation sounded really goofy and almost condescending, but before that I tried to phrase it a couple of different ways and it kept coming out jumbled.