Increasing copyright lengths, automatic copyright coverage, and abolition of renewal requirements have led to the growth of a problem called "orphan works." These are works which may be under copyright, but whose owner is impossible to identify or locate. It's possible that I've violated copyright on orphan works myself; my filk tape of a decade ago, Shrink Wrap Blues, included my musical settings of a couple of pieces by Berton Braley, who wrote large numbers of poems for periodicals and is largely forgotten today. Linda Tania Abrams, who edited a Braley compilation called Virtues in Verse, assured me there were no copyright issues, but it's possible that she was wrong. No lawyer has ever complained to me, perhaps just because its hundred-copy run never got noticed; and as far as I know, no one ever went after Linda. But if we had been wrong and a lawyer pursued it, we could have been up for serious legal damages.
If a work may be under copyright, but you can't find the copyright holder to get any permissions, it may disappear because the existing copies are disintegrating and no one can copy them. Sometimes libraries have gone to considerable expense to restore old publications that were printed on cheap paper, because there's no one to authorize making new copies. The American Library Association has a page on the orphan works issue, with many useful links. On a more personal level, people may want to scan and publish old photographs on the Web, only to realize that they're copyrighted by a professional photographer who went out of business in 1980. When a copyright holder dies as a person or business entity, the copyright passes into someone else's hands, perhaps someone who doesn't know or case.
Because of these problems, there have been efforts to change copyright law to allow a good-faith defense for the publication of works where the copyright holder just couldn't be found. In such cases, ideally, the republisher would be liable only for a standard scale of fees, not for punitive damages, and the matter would usually be handled out of court.
To be fair to all concerned, it's necessary to make sure the good-faith effort really was an adequate one; otherwise unscrupulous publishers could reissue copyrighted works and just pay royalties when they were caught. The problem is particularly serious with photographs, which by their nature rarely come with identifying information. Modern digital images may have metadata identifying their source, but even there it's easy to lose metadata when converting images from one format to another. Copyrighted works don't have to be registered, so there's no convenient single place to look. (And even if there were, the technology doesn't exist to find a matching image efficiently from a very large collection of images.)
A recent article on the Cato site by Timothy Lee discusses these issues and some proposed legislation to address them. There's quite a bit of misinformation circulating (see, for instance, this blog post) about these bills -- claims that anyone using a work can just claim to have made a "reasonable search," and that if a court accepts that claim, the copyright holder can do nothing. In fact, the bills set specific requirements for searches, and copyright holders can still collect "reasonable compensation," just not punitive damages.
The rumor that the bills somehow "threaten open source" is nonsense. If you didn't put your name on the open source code which you wrote and no one can find you, you're up the creek even under existing law.