You raise a good point, Jubilee. We haven't quite settled our approach to copyright rules, though we're having some discussions behind the scenes among the major uploaders here.
I get a threatening or litigious e-mail about once month or two. Thus far I have always been able to satisfy them or fight them off, but my nightmare is that someone will one day succeed in dragging me to court, a cost that would be very difficult for me. Thus the guiding principle is that our approach needs to be "safe" enough to never endanger The Athenaeum. It is frustrating to see many web sites, large and small, seeming to disregard copyright and get away with it, knowing that we have to hold to a higher standard. This will only be more true as we grow in visitors and importance, and as we one day begin to court more museums and art professionals to participate in the site.
I'm not sure we differ so much from Wikipedia. We (as they) apply US copyright law, and as you say, there is a rule that artworks "published" prior to 1923 are public domain, even if the artist died way after that. Another rule that comes into play is that if an artist died more than 70 years ago, all of their works are public domain by default. It's more complex than that, but let's stick to those rules for this discussion.
So let's take Matisse, whose works we are adding to at the moment. He died on November 3, 1954. By the "70 year rule," all of his art will fall into the public domain on November 4, 2024. However, he has plenty of works painted before 1923. If any of those were published, they should be in the public domain, per the 1923 publication rule.
But how do we know they were "published," eh? Publishing as regards art and copyright is typically taken to mean dissemination of multiple authorised printed copies - in books, or postcards, etc. Selling the artwork, or a gallery show (which is "ephemeral" and not a "publication") don't count. If we wanted to be absolutely certain that one of those pre-1923 works was public domain, we'd have to know:
A) That it was published in a book or on postcards or calendars etc before 1923, AND
B) That the publication itself was legal and authorised.
We could (and I think eventually will) build a way for members to crowd-source this, by finding pre-1923 books, going through them, and noting on individual artwork pages that they were published in a certain book on a certain date and are therefore public domain. But that's a lot of work with very little return - how many people will dedicate themselves to doing that? Not many.
Instead, we have to have some "best guess" rules here, and they have to balance respect of copyright (both on its own merits and to safeguard our site) with the desire to make more artwork available to our viewers and members. Right now the rule is the simple one about when an artist died. All artists who died more than 70 years ago should be shown as public domain here on our site. But the class of "artworks before 1923, by artists who died more recently than 70 years ago" is still listed as copyrighted.
I am considering whether we can reasonably apply a different rule, such as "this artwork is older than 1913, so we assume that it was most likely published before 1923". Perhaps coupled with prominent text saying that viewers reuse such artworks at their own risk, and we welcome corrections by copyright holders, that would be enough, and it would "unlock" a lot of artworks.
One thing you should know is that we always keep the full-sized image in the database, and we are planning on unlocking those full versions every year as the artists who died 70 years ago advances by one year. That means that one January 1 of each year we can have a sort of celebration as hundreds more artworks pass into the public domain.
It's an interesting conversation to have, and of central importance to efforts like ours. I welcome your thoughts!