Problem:
Most of the things that come up in Google and Google Scholar for this question have a mostly/completely speculative tone, or at best use models instead of measurements. This report describes some actual measurements in pollen transfer to neighboring fields from GM fields, though they all seem to have been one-off measurements at some unknown time after the GM plants appeared, rather than investigations on if the genes are becoming more common over time, which might imply the ability to become unavoidably common in some environments or even a weed/pest/parasite. This doesn't seem like an obvious problem to me, because the traits associated with those genes are always meant for the context of being farmed or raised by humans and therefore probably wouldn't contribute more to fitness in wild populations relative to the wild alternatives of those genes, but a few counterexamples would smash that belief to bits.
Please explain with the help of example.