The alerts don't know which phones they're going to, they're just shoved out on a mast by mast basis.
While I don’t understand why people might want to turn them off, depending on exactly what you mean (“knows” isn’t really a word associated with tech), I don’t think that is quite correct.
SFIU, and I’m open to correction, a “send alert” message goes from “HQ” to the cells concerned, and each cell then automatically forwards the alert to every device logged onto the cell. The cell has to “know” which devices to address the alerts to - it’s not like a broadcast radio system, because the data have to be addressed.
a record of which devices were logged onto each cell at the time will exist as long as it is retained, and a record of which devices were addressed, but to what degree the device reports back receipt (almost certainly does to some degree as part of “handshaking’) I don’t know, and whether any info about settings within the device are reported back I have no idea, but suspect none.
Someone more knowledgable maybe able to correct or supplement.
[I’ve now found the ETSI standard for the system, but I’m still struggling to understand exactly how it is implemented at cell level …. Whether it does require addressing, or not, and how the alert area is bounded if it doesn’t.]