It'd also be kinda neat if you could somehow tag your trust conditional on a content hash of the service's ToC or Privacy Policy.
It'd be rather impractical in today's world, although maybe machine readable policy formats would make matching easier additions/deletions of key items rather than just a text hash.
Balancing the user effort of reading and flagging each policy term (and considering all the weird implications of each change) vs the utility of being notified would be tricky though.
Maybe some sort of collaborative/curated lists of 'substantial T&C changes' similar to adblocker filterlists could be produced? (Or maybe they already are?)