That sounds like irrational threat-modeling based on paranoia: you're so focused on one concern (abuse by a closed-source vendor) that you're accepting routine

That sounds like irrational threat-modeling based on paranoia: you're so focused on one concern (abuse by a closed-source vendor) that you're accepting routine abuse by other vendors to avoid the theoretical abuse. Your concern seems like a security negative. Also, Apple has demonstrated a great deal of care in the statistics we know they collect, eg by making it impossible to determine your actual input while collecting aggregate statistics. So it is a theoretical concern, but in any practical threat model, it's also a better concern to have than the alternatives.