[TJSomething]

I'm not an expert, but if it's a chip that just synthesizes DNA from a sequence of base pairs, isn't what you're asking similar to making a computer that can't be used to perform evil? I suspect that computing if a given sequence is usable in a pathogen is equivalent to the halting problem. And practically, it seems that a lot of computational resources are required to figure out what a protein does, even for common cases.

[koeng]

If we had really smart software engineers, well paid red-teams, and robust government policy collaboration with industry, I think we can make it at least 95% harder to create something dangerous. We have none of those, though.

It's super hard, but until novel bioweapons are discovered, it is at least a tractable problem.