This page should be active
Yes I know but I cannot activate "szy".
Ask this to the support and wait for a site admin to disable "ais" and activate "szy" (after reimporting the existing translations from "ais" ?). I cannot help at all about this. The split of "ais" in ISO 639-3 is recent (January 2019), but translations were made here before.
Note that "ais" (Greater Amis) is actually not "deprecated" in ISO 639 as you state, but was requalified as "Naotoran Amis" (still part of ISO 639-3) but not as a language subfamily (which would put it into 639-5).
Greater Amis (what "ais" was used for) has now two languages and is a subfamily, where as "Naotoran Amis" (what "ais" is now) is one of the dialects of "ami".
If we interpret "ais" a macrolanguage, translations made in either Naotoran Amis or Sakizaya remain usable in both languages at least as fallback, even if the two languages are hardly mutually intelligible (the same applies for other fallbacks).
Notably Sakizaya will likely fallback to Naotoran Amis, i.e. the northern dialect of Amis. This means that the code "ais" should have been used only for Amis and should have never been named "Sakizaya" on this wiki or in Wikimedia.
In fact I suspect that existing translations in "Sakizaya" made under "ais" are already in "Naotoran Amis" (still vivide) and not in "Sakizaya" (which is moribund, and has most probably largely shifted either Naotoran Amis, or to the dominant Mandarin).
But it would be clearer if some native in Eastern Taiwan could sort these translations: "ais" should be imported by default into "ami", then confirmed by actual translators (if this does not match, they should move the translation to "szy" before replacing or deleting it; but because "szy" is still not active, these translations have to be moved to "ais"; later "ais" will be imported automatically to "szy" once it is open, and all translations in "ais" should be deleted, without redirection).
For now "ais" should not be deleted, because this is the only code for which translations in Sakizaya can be entered (translations in Amis can already be done in "ami").
I managed to create the infrastructure and update the warning banners in all relevant categories. As well I made the cleanup for other active (non extinct) Formosan languages.
Note that you asked the addition of 4 other Formosan languages which were accepted and opened for you, but you did not even managed to contribute to them. So they are left orphaned now. You request for Sakizayan (under "szy") is then likely to fail, until there's another actual Taiwanese translator coming in (and that's why I better prepared the way to allow them to come more easily; they will probably come here after visiting the Wikipedia incubators to improve the test support, and then interest some Taiwanese linguists, or students/searchers in Taiwanese cultures, or some Taiwanese governmental or educational worker, or some local public offices and medias that want to better support these languages with the help of their local communities).
For the record…
Regarding the "deprecated" status of "ais", Amir is almost certainly alluding to the fact that SIL considers the code to be "Deprecated" following this change request (of December 2018). As they explain at their website, such deprecated codes "are no longer in active use, but still have the meaning assigned them when they were established in the Standard."
As to whether "existing translations in 'Sakizaya' made under 'ais' are already in 'Naotoran Amis'", I cannot personally say, but User:StevenJ81 (Langcom member) and User:Corainn (zh-Hant-TW-N user) have discussed this and seem to agree (each from their own perspective) that this is not the case.
In any event, as stated by StevenJ81 in the same discussion, this issue is currently being worked on in phab:T174601. Verdy_p, if you do not agree with the action plan laid out at phab:T174601#5348652, you should comment there.
Edit: StevenJ81 is technically a "clerk" (a non-voting member) of Langcom.
So my comment is still valid: We should have not named "ais" as "Sakizaya" given that "ais" was given the meaning of "Naotoran Amis" in ISO (not SIL which is just the current maintenance RA and does not decide itself!), which is in fact a local dialect of "Amis" but not "Sakizaya" for which ISO assigned a distinct code. ISO also assigne the code "ami" for "Amis" (from which "Naotoran" is a village where both languages "Sakizaya" and "Amis" were used by the two coexisting local communities, and that's why "ais" was named "Naotoran Amis" and not just "Naotoran" which would really have been confusive).
So there was no error in ISO. But the deprecation of "ais" just means that ISO no longer considers it as a distinct language, but as a dialect of "ami", and what occured in ISO was actually not a "code split", but a requalification to cover "Amis" more extensively and "Sakizaya" was never encoded and covered by ISO. So this is another case where an ISO 639-3 code is assigned to a dialect, even if it is not in scope (this can be applied to Maltese [mt] and various other Arabic dialects that also have their own ISO 639-3 code, even if they are normally not in scope of ISO 639-3 which is not supposed to cover dialects, but we also know that the encoding of dialects in ISO 639 was tried and published in a early 1st version, and rapidly abandonned).
We should have never used the name "Sakizaya" in Wikimedia for "ais", as it was wrong since the beginning (even before the ISO change in 2019). It's up to us to not use "Sakizaya" for "ais" and to make it clear that this was an error of Wikimedia (and Translatewiki.net) and not from ISO.