Many of them look like they are out of copyright anyway, although not all.
Ever since Bridgeman (http://en.wikipedia.org/wiki/Bridgeman_Art_Library_v._Corel_...) most museums have accepted that no matter how much effort goes into the imaging, some imaging simply does not qualify as novel enough to warrant new copyright. This ruling is, as I understand it, not actually binding in other districts, but it's generally accepted.
Because imaging of flat art, without a frame, is inherently intended to add as little interpretation as possible, it's generally accepted that reproduction imaging of 2D art is not copyrightable. This is the Getty's own position (I worked there until a few months ago, from the same department that brought this project out).
Any other questions I can answer about this project I will.
Disclosure: I work at the Metropolitan Museum of Art, and am trying my hardest to share all our data with the world, but have learned that what is legal and what is museum policy can and often do clash. Its super interesting.
If you want relatively low resolution images of our collection, feel free to visit http://scrAPI.org , which just scrapes our live site. Its a workaround until our policy changes.
Roger, how was working for Getty while you were there? Was this your main project? How supportive where those involved in sharing the data? If hackernews isn't the most appropriate place to answer these questions, I'd be glad if you could take the time to go off-site for a conversation.