Several commenters asked: If AI companies do not own the copyright in the works created by their programs, how can they be held liable under defamation law?
Defamation law and copyright law are two different bodies of law, aimed at serving different interests and with different definitions. Because of this, people can routinely be sued for defamation even when their works are not copyrighted. For example,
[1.] The phrase “John Smith is a convicted child abuser” is a short and simple phrase that is not protected by copyright. (Even if some very creative short phrases may be copyrightable, the combination of an already existing name and the existing phrase “is a convicted child abuser” is certainly not copyrightable.) Still, it could indeed be defamatory.
[2.] If Alan the Author writes a defamation of Paula Plaintiff (pro tip: never defame people whose names start with P), and Donna the Defendant copies Alan’s defamation, it may indeed be defamatory, if Donna has the requisite mental state—even though the copyright is owned by Alana, not from Donna. (This is true regardless of whether Donna copied Alan’s work with his permission, engaged in fair use, or infringed Alan’s work.) This is actually often the case when Donna is a newspaper publisher publishing Alan’s copyright. (Newspapers often publish comments from people who are not their employees, without obtaining an assignment of copyright, but only a nonexclusive license to publish; in that situation, the copyright remains with the author, but the newspaper may be independently liable for defamation.)
[3.] If I spontaneously say, “John Smith is a convicted child abuser,” and add five sentences of explanation, that verbal statement is not copyrightable because it is not fixed in a tangible medium of expression. But it could be defamatory, assuming the elements of defamation are satisfied.
In all these examples, the defendants are potentially liable, because the defamation law takes into account what the defendants communicated and whether it was false, damaging to reputation, uttered in a necessary mental state and unprivileged. But the defendants are not copyright owners, because copyright law is concerned with a very different matter (providing an incentive for creative expression fixed on a tangible medium). This in itself does not tell us that AI companies are responsible for the operation of their copyright-free programs, of course; I discuss this in much more detail here. But I think it explains why copyright law has nothing to do with the matter.