I get where you're coming from.

We've been able to make new viruses for decades.

Scientists were DNA editing together boutique polio strains way back in 2002.

This news article is about is a machine learning model which you feed bacterial DNA, and it (quite successfully!) generates a genome for a virus that can potentially kill that bacteria.

Pretty much every scientific advance has the potential to be used as a weapon. These scientists have taken very weaponisable technologies technologies (deep learning, virus synthesisation) and used them to create a new technique that can cure people suffering from late stage and antibiotic resistant bacterial infections. Could another research lab try and use a similar technique to develop bioweapons? Sure, maybe. That's not what's been done here, and this work hasn't pushed the frontiers of those technologies in a particularly weaponisable direction. I'm very convinced that their work specifically is going to save far more lives than it takes.