I mean, if you guys just want a pile of trash and we call it a derelict, I can certainly do that. If I don't make the trash pile have a bunch of gaps in it, you'll still be able to see it from afar just fine.
As far as any objections to particular believability bits: frankly Keith and I historically tend to think that technobabble can sort out most things.
1. Why doesn't the AI realize the benefit the players are getting from the derelicts and do it themselves? Because they don't rely on optical sensors, perhaps (why would they?), and there's something only apparent in the visual spectrum. Or because they already have a much better source of resources. Heck, maybe they ARE using them.
2. Why not destroy them if they know what they are? Maybe that would cause a really really big explosion. Or maybe they just figure they'll recapture them soon. If someone invades your own country, generally you don't firebomb your own fields and factories unless you REALLY don't think you can get them back. Sabotage them, though, sure: and we already see the AI doing that. If the AI assumes it's going to win, its behavior seems entirely reasonable to me.
3. Isn't advanced technology risky there? Sure, but perhaps these are things that accidentally happened when AI weapons of some specific sort hit a certain kind of human propulsion system, creating these too-small-to-do-much gates between universes. All you can really do is suck matter through the cheese grater that are these existing gates. This would also explain why you can't move these things. So nobody made it on purpose, and it's not useful in anything remotely close to its current form in general.
Or pick your alternate BS explanation. Basically anything you can come up with that is a problem, things can be twisted to make that not a problem. That can then create new problems, but then just twist and repeat until there are no problems left. And now you have Star Trek!