Not so much creating as packaging. In one of the articles you linked to speaks about, for example:
"If common medications for chronic diseases were available in the 3-D printer, a customized “polypill” could be created that could potentially contain all the medications a patient needs in 1 pill. Think of how much easier it would be for a patient to take 1 pill daily instead of 15, of how much this could improve adherence."
So, more about dispensing than creating drugs.
Actually, we're kinda talking the same thing. A polypill is a combination of drugs in one gelcap, capsule, solid pill (or other forms). There are a few in the market today.
To create one with a 3D printer, you'd have to combine the pharmaceuticals, binders and whatnot according to a formula, just like a standard pill today. So in the sense that it is packaging things from a known formula, it is sort of "dispensing".
But it is doing so in individually customized pills- something we can't really do today. Not with any efficiency, that is. And since it would be making pills that at pharmacologically unique, it would be "creating drugs".
Ditto the case in which they're talking about tailoring things like chemotherapy or gene therapy.
And another speaker also notes:
"Meeting the regulatory requirements of the US Food and Drug Administration could be a hurdle to be cleared before large-scale use of 3-D printed products could be realized. We are talking about an immensely complex regulation process. Different manufacturing regulations and state board requirements could impose obstacles to the adoption of the drug printers in practice. An imperative difference must be established to distinguish drug printers as manufacturing or compounding technologies."
That's all a given. Of course there would be regulatory hurdles to jump. And the EU and other major countries would have their own regulations as well.
For the illicit trade - so they can make what they have in the form of a pill. Yippee.
Well, yeah.
They already make their own pills. If the 3D tech is compact and clean enough, it would make detecting their manufacturing sites much harder to find. If pharma-printers become able to work from downloadable formulas or data on a removable drive and the right mix of ingredients, the task become harder still.
For instance, if the Hell's Angels want to make a meth lab today, one hurdle they face is that it is currently much harder to acquire sudafed than it was 30 years ago. But if your pharma-printer can make sudafed- or just the ingredients you need from it- with substances that aren't scrutinized the same way sudafed is, meth labs get that much harder to find.
It is potentially like how, before OKC and a couple of other events, certain fertilizers could be ordered with impunity. But now, ammonium nitrate fertilizer is not only scrutinized, each maker includes microscopic beads that are unique identifiers for the manufacturers.
...except you might not be able to include analogous tracing techniques in a pharma-printer, since that could potentially screw u the machine and/or the efficacy of the product.