I would presume that, whatever they say, the real reason for that is account deduplication.
To block spammers in a social network, you either need messages to be expensive to send (ala the old e-stamps concept) or you need 1. identities to be expensive to acquire, plus 2. the ability to ban identities. The simplest way to make identities expensive is to require that each account be tied to some sort of real-world scarce token that you can prove possession of, like a phone number.
If, when you ban a user, you also ban their token from being used to create new accounts, most casual users are stymied from "re-making", and professional spammers have their potential volume (and therefore potential ROI) lowered by an order of magnitude because they need to pay for phone numbers (at e.g. $1 a pop on Twilio) to in order to register the spam-accounts, and each of those spam-accounts (and therefore phone numbers) will be banned quickly enough that they won't have made $1 back by the time it happens.
The fact that account-deduplicating like this helps them get more accurate active-subscriber statistics is a nice bonus.
What this probably isn't, is a KYC measure. They don't care who you are; they just care that—whoever you might be—you only have one active account. (If you can think of a better way to achieve that without asking for potentially-identifying information, I'm all ears! Something something proof-of-stake? Make registration require you to burn an hour/day/week mining a token?)