That's not uncommon in API design and is something I've even built into APIs I'd created before because... yeah, sometimes you want to recover data that was "deleted." Not sure how that meshes with privacy laws these days though.
I once had a stint of 'all but Sysop' on a BBS, because I wasn't as busy IRL as the actual Sysop/board-coder and was amongst the more active of the existing Moderators.
Messages on the wrong sub-board could be moved by moderators with a suitable jurisdiction (or the author), with explanatory message, and one of the places they could be moved was "Limbo", a write-only (even to moderators) board for anything deemed not in any way redeemable (everything from low quality -
something I fell foul of on more than one occasion - to approaching/actual illegality -
I think I avoided this issue - as well as personal regret). And so, for a while, I
could read what was there (though Limbo had a move-to-delete option, IIRC, so no doubt there were things that never survived to be seen by me), and of course I could have de-Limboed things.
To lack an equivalent feature (i.e. a last resort of a complete /dev/nul 'destination' only) would be more problematic than having it for management reasons, including author-'deleted' items, because errors (or maliciousness by A.N.Other having gained access to your login) always happen.
This was well before the Windows Recycle Bin (though Mac's trashcan was certainly already a thing) and UNDELETE in DOS was the closest most people had to reversibility of such things (didn't work if you'd RDed the directory it had been in) unless you got into Norton or precursor tools. But even these days, or 'especially' given the vapourish and sometimes literally cloudlike nature of data, there's a dichotomy in security in that users expect
both that nobody should access things that they don't want accessed (by anyone, ever)
and that if they lose access to things they never meant to lose that there's a practical way to get it back. The fuzzy buffer between the two can only be maintained by a User Service Level Agreement that those who
can bridge either gap between publishable and remnant data may and will only do so upon sufficient (without being impossibly onorous) 'permission' to do so.
Unless you're specifically designing a transient-only storage/viewing system (for one of several possible/combinable reasons specifically to address provider and user experiences), the clientele
must accept this (certainly must have 'signed' the many pages of legal mumbo jumbo that says as such).
I'm not surprised that any amount of information is there. Unless they do as much as anon.penet.fi's final act to save its users from the rabid scientologists, purposefully or otherwise, it'll stay available for the right circumstances.
And allowing that archived data to be retrieved by clients is a different matter completely, and definitely shows some negligence on their part.
Hooo yeah... That was a booboo (for someone). Not necessarily the person who enabled it, if they had that intention.
But at least it's entertaining to hear about. I've not got my ear as close to the black/grey/white-hat hackers as I once did. And I suspect these were 'shiny' black-hats, actually glowing with righteousness in the right light.