I'm a little confused. I definitely think there are safety issues with what I posted (see my disclaimer at the top), but not the ones you've mentioned. I think you might be misinterpreting the setup.
That's true, but mitigated somewhat because the shaft is supported at the bottom where it runs through the hole in the drill table, so there's no real torque being put on the bearings. There IS a straight sideways force, but a hell of a lot less than the thickness-sander conversions (for scale-making) I saw elsewhere. Also the sanding seems to work best with minimal pressure. I could fix it entirely by raising the table to maximum height (providing support at the top of the shaft just below the chuck) and supporting the shaft bottom at the drill base.
Do you mean the shaft itself? The shaft is wooden. The only metal in this setup is the blade itself but I can't really picture a way the shaft could break that would throw the blade. Even if the shaft broke, one half would still be in the chuck and the other half running through the table.
I'm not sure what you mean here. The blade never touches the baseplate or any part of the drill. I hold the blade by hand against the sandpaper drum which is spinning in the drill. That's actually the dangerous part, IMO, because my fingers are so close to the spindle and holding the blade. I've had the blade lurch in my grip LOTS of times so far without any damage, but from a safety point of view cuts are inevitable and protective gloves are out of the question so close to the spindle.
I made a magnetic jig, too, for using with a dremel, but I found five seconds with the dremel more worrying than 5 hours with the drill press rig. I may try using that jig to hold the blade against the drum to get my fingers out of harm's way, but I'll probably add another magnet, first, for extra holding strength.
Thanks for your criticism, although it's not just a case of being too cheap to buy a flex-shaft because that's not how the setup works. If anything it's more like avoiding a buffer/grinder.