Viral reverse transcription takes place in a semi-permeable environment that is accessible to small molecules (e.g. dNTPs), but not to macromolecules such as proteins or nucleic acids. Cellular RNAs cannot access reverse transcription complexes. Furthermore, RT uses a primer to initiate reverse transcription: retroviral genomic RNA comes with a primer annealed to it (tRNALys3 in the case of HIV-1)
It is probably because RT forms part of a specific ribonucleoprotein, which is the PIC by means of protein-protein and protein-RNA interactions. Moreover, the number of RTs per infected cells in not that high, so I can imagine that, even free in the cytoplasm, the polymerase would affect very few mRNAs. Finally, 5' and 3' sequences within retroviral RNA play an important role in Reverse transcription and they are not present in mRNAs. These are the explanations that I find logic.
Viral reverse transcription takes place in a semi-permeable environment that is accessible to small molecules (e.g. dNTPs), but not to macromolecules such as proteins or nucleic acids. Cellular RNAs cannot access reverse transcription complexes. Furthermore, RT uses a primer to initiate reverse transcription: retroviral genomic RNA comes with a primer annealed to it (tRNALys3 in the case of HIV-1)
A bit off-topic but in very rare instances cellular RNAs can be reverse transcribed. At least that's the model for the src gene in Rous Sarcoma virus. There are other examples as well.
I believe that some cells are affected. In general, this could be because an interaction between cells protein and the RNA-DNA transcriptase. Not sure.