All genomes exhibit nucleotide composition biases; for example, the dinucleotide CpG is dramatically underrepresented in vertebrate genomes. RNA viruses that infect vertebrates, such as influenza A virus (IAV), mimic this CpG repression. This mimicry is partly due to evolutionary adaptation to subvert detection by host factors like the zinc-finger antiviral protein (ZAP), a cytoplasmic sensor that specifically targets CpGs in single-stranded RNA. Prior work from our lab enriched for CpGs in two IAV genome segments (1 and 5, both encoding components of the viral ribonucleoprotein), uncovering segment-specific variation in the impact on viral fitness. Expanding on this, here we have produced IAVs with CpG-enrichment in genome segments 2, 3, 4, and 6. The greatest attenuation resulted from CpG-enrichment of segments 1, 2, and 3, which encode the PB2, PB1, and PA subunits of the viral polymerase. This attenuation was ZAP-dependent, as viral fitness was restored in ZAP knockout cells. In contrast, CpG-enrichment of segments 5 (nucleoprotein, NP) and 6 (neuraminidase, NA) was well tolerated. Triple mutants combining CpG-enriched segments 1, 2, and 3 showed an additive attenuation, which was also rescuable in ZAP knockout cells. This work suggests that the polymerase segments are the optimal targets for CpG-enrichment in live-attenuated vaccine design and highlights the potential of this approach. Ongoing research seeks to elucidate how ZAP-RNA binding impacts viral RNA transcript abundance and structure for viral mutants that are attenuated by CpG enrichment compared with those that are not, which will enable refined design of CpG-enriched virus vaccines.