This is a great question that I don’t know the answer to. If someone was to happen upon a set of seed words, I wouldn’t think that they would be able to know that they were associated with a multisig setup but I don’t know for sure.
Discussion
That's a good line of thinking. I'm pretty sure the best they could do is generate xpubs for potential derivation paths for that set of seed words; then, if they also happened across a multi-sig wallet descriptor backup that contained a matching xpub as one of its keys, they could then conclude that the set of seed words was indeed used as part of that particular multi-sig wallet
BTW, since a 2-of-2 would absolutely require both keys (by a single user in this case) to be useful, there wouldn't be as much reason to save a wallet descriptor backup (that someone else might happen upon) as you probably would/should for a 2-of-3, etc,
Sounds legit to me. There would be advantages to this over just a single sig. Even if the one set of words wasn’t as secure,like you mentioned, distributed among multiple places(easy to access and not lose) the other key could be secured as one would do as a typical single sig. This setup could have advantages over a single sig setup but not as robust as a 2 of 3. Food for thought anyways, brain food.
It also occurred to me that you could easily have many 2-of-2 wallets - all with same same secure seed combined with a different 'insecure' seed.
Though, I believe this would allow for block analysis that could link outputs to addresses of different 2-of-2 wallets that were reusing the same signing address from the the secure seed - but, I'm really starting to getting beyond my knowledge here.