In computer programming, we all know that every time you write code, you introduce new bugs. As a non-biologist, my mental image of genetic modification (of the newfangled variety, not the old style where cross-breeding and selective breeding were involved) is that it has the potential to add new bugs into the code that we won't understand until we "run" the code -- ie have it in a living organism inside the context of the biosphere -- and that the problem with that is that if we create a bug in the code that has detrimental aspects in so far as our interests are concerned along with useful aspects so far as the survival of the new piece of code is concerned, we can end up with some serious problems down the line.
When we write software, we test it inside of a sandbox before we push it up closer to the real-world scenario it was meant for. In some cases there are many levels of bug testing between the developer's sandbox and the end user's machine, and yet still, we get bugs in live code. If that live code then went around breeding, spreading those bugs before the developer had a chance to fix the error, that would be a serious problem. Unintentional malware, so to speak.
The Cheshire cats of Windup Girl by Paolo Baccigalupi are a great fictional example of what I mean. Cute idea, but the Cheshires were better adapted than ordinary housecats, and they out compete ordinary housecats into extinction in the fictional world of the novel. That sort of thing happens in real life when we bring a plant or animal species from one geographic area into another area that was previously protected from that species. In some cases the "invasive species" takes over and the "native species" dies out very rapidly.
That is what makes me nervous about GMO anything.