We clearly need more #ResPublicae posts/media to be marked as sensitive. I've taken action on one account. Later I'll survey the most followed accounts, see if this needs to be done instance-wide or on a case by case basis.
(Thanks for reporting graphic images of animal abuse, blood!)
Other possibilities exist. The principle of least astonishment could be a guiding factor:
https://en.wikipedia.org/wiki/Principle_of_least_astonishment
We could mark posts from certain accounts unlisted if they regularly post to hashtags in ways which don't belong.
We could also forcefully add CWs to posts, but so far we don't. It's hard to come up with a CW which would be applicable for all accounts, or for entire classes of accounts equally. A catch-all CW like "From birdsite 🐦" is broadly correct but also potentially unhelpful.
Some accounts with no followers are sometimes considered problematic too, even though they're hard to find. Someone may accidentally stumble on something they were not prepared for.
I've surveyed a few accounts from known-controversial quarters and I've marked a few as forced-sensitive. I've also disabled mirroring RTs where they're mostly not about EU policy.