I'm recently listening this replacement a lot on youtube, it's as if the practice is on the rising. Is it? what conditions such occurrence? Where in the US is it happening? And how did it all begin?
This retraction happens when [s] is in a consonant cluster with a following [ɹ] (as in the video you linked, "just straight up"). The reason is articulatory: the retroflex gesture of the [ɹ] is anticipated, so the [s] becomes retroflex/postalveolar too. It's an assimilation. This is pretty common in US English: I associate it with Southern and African-American speakers.
Without specific examples, it's hard to say what exactly you're hearing.
Different languages (or accents) have different boundaries for what exactly counts as /s/ vs. /ʃ/, so it's possible that the speakers you are hearing do not really "replace" their /s/ with [ʃ], but just use a pronunciation that sounds more like [ʃ] to you.
But it's also possible that they really are using [ʃ] (possibly even /ʃ/) where a typical reference English accent would have /s/. It is known that English /s/ can have retracted pronunciations; I haven't personally read a lot of the literature about this topic, but there seems to be a decent amount (maybe some of the references here would be helpful?: Variability in American English s-retraction suggests a solution to the actuation problem).
Sometimes this happens in environments where the contrast is neutralized by English phonotactics; the most well-known examples I think are the word-initial consonant clusters /str/ (as TKR says) and /stj/ or /stʃ/ (the cluster at the start of "student" in a typical southern British English accent).
Glain (2013) says that /st/ followed by a vowel, as in stop and start, also has a variant pronunciation that sounds like [ʃt]. I don't know that much about this, but this indicates that s-retraction is not restricted to the context of str-, although it may have started out there and subsequently spread to certain other s-initial clusters. I haven't heard of s-retraction occurring outside of consonant clusters.
German Phonetics and Phonology: Theory and Practice (By Mary Grantham O'Brien and Sarah M. B. Fagan, 2016), says
[s-retraction] has in fact been documented in the English of speakers from the United States, Great Britain, and New Zealand (Rutter 2011:27). It appears to be spreading, not only by being characteristic of the speech of more and more speakers, but also in the environments where it occurs. Janda and Joseph (2003:212) have documented the [ʃ] pronunciation of /s/ in a number of environments other than the sequence /str/:
(6) [ʃ] in English clusters other than /str/:
a. with r preceding: under[ʃ]tand, thunder[ʃ]storm
b. before [r] alone: di[ʃ]respect
c. before a plosive other than /t/, with /r/ following it: [ʃ]prinkler, [ʃ]creen
d. before a plosive, with no /r/ involved: [ʃ]till, [ʃ]chool