I think you might be over estimating how effective these technologies are. Even in the most controlled circumstances, the amount of actual specific editing which occurs is probably quite low. Off target edits are likely also occuring. The focus for gene editing currently is on terminal and generally low quality of life diseases, as the low efficacy and high risk are more acceptable. We are many years out from robust and stable editing using a delivery system like this in an ideal clinic environment. Weaponizing this technology in any meaningful way is not something we probably will have to worry about in our life time. The paper is not open access so we can't know for sure what exactly they did, but it is in mice and seems to have a florescant read out, which suggests a rather crude edit (did we chop up the gene enough to turn the lights off).

We have things in the air all around us that are small, stay floating in droplets well, contagious, and self-replicating (viruses).

We know how to make things that are lethal in incredibly small doses.

Compare to this, which is not likely to stay in the air well, not self-replicating, and not-contagious, and would require larger doses for lethality.

Sure, we should be careful with any new technology, but the options to weaponize this or to have unexpected bystander externalities seems really small compared to other established technologies and risks we bear.