The first red flag that should leap out at you, but perhaps only if you have some knowledge of virology to begin with, is the very idea that you could "know" what certain mutations would do — i.e., that you'd know what you'd create and what effect it would have on humans — without extensive testing in humans themselves.
For COVID-19, for example, we know that:
among the infected, asymptomatic cases represent a significant percentage of total cases,
while severe cases are relatively rare: 14% of total cases,
and critical cases, where respiratory failure, shock, or multi-organ dysfunction occurs, represent just ~5% of total cases,
and that only 2.3% of total cases are fatal, with every single documented fatality occurring among the critical case population.
This means, right off the bat, that if SARS-CoV-2 were engineered for the purpose of infecting and severely harming humans, it would have had to have been tested in at least hundreds of human subjects in order for scientists to know how effective it was. While we do have the ability to manipulate the genomes of viruses, or any other organism, for that matter, what we don't have the ability to do is to know how that will translate into effects of the virus in human (or any living) subjects.
