It's interesting in this case (and probably many others) how there is a direct tradeoff between device security and data privacy.
If everything is generated externally and only a preview image is sent to your device, there is no security risk to your device (unless you open the link), but a privacy disadvantage.
If everything is generated on-device, there's no privacy issue in terms of third party services, but there is a privacy issue if the link is being used maliciously to track the user, and there's a potential security risk if there's a vulnerability on the page that requires no user interaction.
Of course, on the privacy side, if any sensitive content being linked to doesn't require a login, then it is only offering security by obscurity, which is so bad from a security standpoint already, so that's kind of a moot point. You likewise shouldn't be pushing passwords or whatnot in the URL.
Which is to say the researchers are right that the potential privacy hit is better than the potential local security hit, although I'm loathe to say that when Facebook is involved since you can be pretty sure they're going to use this to abusively harvest and store any user data they possibly can.
I don't see Apple Messages anywhere on that list, and I know it generates previews, so I'm assuming they're the redacted one?
Interestingly, I've noticed that Messages will generate a preview of links from contacts in my address book, but does NOT generate a preview of links from other contacts. So I don't get previews from spam links or things like UPS tracking alerts, but I do get them from friends and co-workers.
This isn't perfect from a security standpoint, but seems like a not-so-bad compromise.