One major controversy I recall with VRchat's implementation of age assurance (via Persona) is that they retained identity of who you were after completing the process. Not who exactly, but your identity was hashed so that if the same person attempted to verify again they could tell who it is. This effectively meant one individual can have one validated account per lifetime. If the flakey AI moderation banned you for an emoji combination freshly deemed racist/etc, you effectively lost your only shot at a validated VRchat account forever.
This is an issue with age assurance that goes well beyond just verifying age, and is undoubtedly viewed as a fringe benefit of age assurance for those wishing to deanonymise the internet. There's obvious problems with this beyond just "can't ban evade anymore". Authorities or intelligence can run bulk hashes against sets of IDs to effectively deanonymise all accounts.
The question here is, does Discord do the same thing for any or all of the selected options? Or is identity data full well and truly purged once the outcome is determined - no hashing, nothing retained other than the result?