There are so many potential mutations that putting specific evolutionary pressure on a virus in order to "prevent" only those specific effects has such low efficacy considering the size of the problem space.
GOF research has provided very little in terms of benefits, and many argue it's simply been a way to disguise bioweapons research after the Geneva convention.
https://en.wikipedia.org/wiki/Lancet_letter_%28COVID-19%29?w...
You don’t have to be a rocket scientist to know that is a terrible idea. “Sticking our head in the sand” appears to not fit this use case very well
In reality COVID-19 required several key mutations to be as harmful and to bind as effectively to human cells as it did. The likelihood of all of these mutations occurring naturally, in animals only, without intermediate variants to observe and develop immunity against, is very small. If only 64 amino acids must mutate exactly correctly to create the deadly variant then it would require 4^64 or 2^128 mutations, which is a large number. Assuming it mutates 100B times per year, that's still more years 3e27 years.
For example, this experiment on more transmissible HIV requires hundreds of acids (600-1000, and a 32 amino prefix).
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2836558/
The point being that creating extremely unlikely events to "see what would happen" is only justified if the risk of the observation is less than the natural risk of the event. If the risk of a world-ending pandemic is 1e-128 per year, and the risk due to human-made viruses is 0.1%, then "not sticking your head in the sand" has raised your risk by an astronomical factor.
EDIT: the evidence in favor of wet market zoonosis is what, exactly? "It's always been zoonosis in the past" plus "we (claim to have) first found this in a wet market" and maybe also "there's some RNA fragments in the wet market samples that show interesting animals plus covid"? As far as I'm aware that's it.