False positives of the kind you're thinking of aren't possible--it's checking for hashes that match known bad images, not running machine learning/image detection to detect if the photo you just took contains bad content. The issue is that there's nothing stopping Apple/the government from marking anything it finds objectionable--like anti-government free speech--as a Bad Image, beyond CSAM.
Additionally, tenants are often unaware of their rights and don't exert them for that reason. Others are aware of their rights but don't have the resources to fight if their landlord violates those rights.