If by “rogue proteins” you are referring to proteins that could be incidentally produced as a result of new cryptic reading frames, my co-authors and I discuss this in a publication (Herman et al., 2011). In brief, proteins could theoretically be produced by “hidden” or “cryptic” DNA sequences. In the case of GM crops, such sequences would be generated at random. The chances of assembling a gene that actually expresses a protein by random is very low, the chances of this protein being functional is miniscule, and the chances of it being a safety concern are astronomically small. Because far fewer random changes occur during the insertion of a GM trait compared with traditional breeding processes, the chances of producing a rogue protein during the development of a GM crop are orders of magnitude lower than the chances of producing a rogue protein during traditional breeding. Our experience with traditional breeding tells us that this is a very, very remote possibility. Even so, and unlike traditionally bred varieties, molecular characterization of transgenic events allow us to say with great confidence that rogue proteins are not produced by the DNA within or adjacent to the GM trait insertion.
For further reading on this topic, I encourage you to read an article by science journalist and blogger Alan Dove at http://alandove.com/content/2013/01/whos-afraid-of-the-big-bad-orf/ where he tackles some of the confusion that has arisen on this topic.
References: Herman, R.A., Song, P., and Zhuang, M. (2011). Safety risks of cryptic reading frames and gene disruption due to crop transgenesis; what are the odds? GM Crops 2, 4-6.