SARS-CoV-2 has a 12-nt-long exogenous sequence into the S gene, encoding the polybasic furin cleavage site, including the CGG-CGG arginine dimer. In this regard, I recently posted two facts: (i) this 12-nt and the reverse complement sequences are identical to a many highly expressed sequences in human cells, suggesting that a progenitor of SARS-CoV-2 jumped into humans and recombination with human cell mRNA produced the pandemic virus; and (ii) the identification of a representative group of SARS-CoV-2 specimens with arginine codon usage bias in the furin site. Are these two independent facts? To address the issue, the aim of this work is to trace the arginine codon bias in the dimers of this amino acid in the furin site from virus updated genomes. At S protein level, the acquisition of the furin site was enormously favourable for the virus. For the humanity was (or is) the new and unprecedented covid pandemic. Moreover, at S gene level the CGG-CGG arginine code was enormously unfavourable for the virus. Since arginine has six codons, the virus can overcome the situation through a codon usage bias. Here I show some SARS-CoV-2 Omicron sub-variants BA.4 and BA.5 (a July, 2022, sample) with furin site arginine dimer codon optimized: 8 out of 4,986 (0.16 %). In SARS-CoV-2 the synonymous base substitutions are not common, however, in this particular case of the arginine, it is the only way that the virus has to stabilize its genome. On the other hand, from science, this arginine codon usage bias can support the hypothesis of recombination for the SARS-CoV-2 polybasic furin cleavage site origin. Finally, this work opens the door to monitor the course of that codon bias in the new virus genomes that are coming.