This is just inherent to compression. The whole idea is to represent many bytes of data with fewer bytes. That means that the bytes you have each tend to count

This is just inherent to compression. The whole idea is to represent many bytes of data with fewer bytes. That means that the bytes you have each tend to count for a lot. So a small corruption has a huge effect. To solve this, the format itself needs to incorporate error correction. There are various ways to do that with different trade-offs, but all of them will hurt your compression ratio, because the only way to correct errors is to have some kind of redundancy, and removing redundancy is precisely what compression is about.