Don't scientists out there very explicitly avoid building self-replicating systems at any significant scale to avoid the associated risks?
I think they already can create artificial/synthetic lifeforms in a lab - it's just that there are not a lot of use cases that make building a factory worth it, at least so far.
[Upd: I stand corrected - I thought they can, but turns out they can't. Thank y'all.]
All of these "synthetic lifeforms" I have seen are usually very gimped and nowhere as robust as a normal bacteria. It is still an ongoing effort to make a real, designed from scratch, comparable bacteria that can match the original. But speaking as someone who worked on these things, I would not expect any major breakthrough soon.
It would be nice to think that a scientist in command of self-replicating artificial life technology will have the restraint to hold back on deploying it at scale. But if someone comes along and says “hey I made an artificial bacterium that can eat all the CO2 in the atmosphere, want me to replicate it globally?”, someone is going to push the button.
The only self-replicating thing we can "build" is a modified biological cell and that’s already a tremendous achievement.
In non-biology, most of the work has been self-assembling, not self-replication. IE, put all the puzzle piecees in a bag, jiggle for long enough, and you get a fully assembled puzzle out.
There's also https://en.wikipedia.org/wiki/Xenobot
Realistically, many scientists are working towards fully self-replicating machines. Generally, nobody has been able to articulate a realistic danger that is not highly implausible; we work under the assumption either nothing bad will happen, or we'll be able to stop it well before it's an issue.