So what would stop the list of hashes from being extended with hashes of copyrighted media, evidence of corruption (labelled slander or an invasion of the perpetrator's privacy) or evidence of the preceding abuses of the system themselves?
Once you have an established mechanism for "fighting crime", "don't use it to fight that type of crime" is not a position that has any chance of prevailing in the political landscape - see also all the cases of national security wiretaps being used against petty druggies.
Hashes don't really work that way. They don't actually give any high level of a photo's contents. You can't ask a hash to find all photos of a certain document or a meeting or anything like that. They really only detect exact copies, which makes them somewhat useful only for the most basic of copyright infringement (i.e. proving someone has a copy)
As far as I remember, Apple's proposal was to involve https://en.m.wikipedia.org/wiki/Perceptual_hashing which is meant to sidestep this exact problem - and either way, your objection would be equally applicable to CSAM. There is no mechanism that works better for it than for copyright enforcement.
> So what would stop the list of hashes from being extended with hashes of copyrighted media, evidence of corruption (labelled slander or an invasion of the perpetrator's privacy) or evidence of the preceding abuses of the system themselves?
Absolutely nothing. But they are already scanning hashes now on everything.
Once you have an established mechanism for "fighting crime", "don't use it to fight that type of crime" is not a position that has any chance of prevailing in the political landscape - see also all the cases of national security wiretaps being used against petty druggies.