September 9, 2024
Fair Use or Foul Play? Free Books Cross the Line
Last week, the U.S. Court of Appeals for the Second Circuit affirmed a federal judge’s March 2023 holding that the Internet Archive’s practice of digitizing library books and making them freely available to readers on a strict one-to-one ratio was not fair use. For reasons I’ll get into below, the outcome is pretty unsurprising. It’s also worth looking at because it likely previews some of the arguments we’ll hear in the case between the New York Times and OpenAI (creators of ChatGPT) and Microsoft if (or when) that case makes it to the Second Circuit. (Quick summary of my post on the subject: The New York Times Company filed suit late in December against Microsoft and several OpenAI affiliates, alleging that by using New York Times content to train its algorithms, the defendants infringed on the media giant’s copyrights, among other things.)
First, some background. The Internet Archive is a not-for-profit organization “building a digital library of Internet sites and other cultural artifacts in digital form” whose “mission is to provide Universal Access to All Knowledge.” To achieve this rather lofty goal, the Archive created its Open Library by scanning printed books in its possession or in the possession of a partner library and lending out one digital copy of a physical book at a time, in a system it dubs Controlled Digital Lending.
Enter COVID-19. During the height of the pandemic, when everyone was stuck at home without much to do, the Archive launched the National Emergency Library. This did away with Controlled Digital Lending and allowed almost unlimited access to each digitized book in its collection.
Not surprisingly, book publishers, who sell electronic copies of books to both individuals and libraries, were not thrilled. Four big-time publishers — Hachette, Penguin Random House, Wiley, and HarperCollins — sued the Internet Archive for copyright infringement, targeting both its National Emergency Library and Open Library as “willful digital piracy on an industrial scale.”
The Internet Archive responded that these projects constituted fair use and, therefore, did not infringe on the publisher’s copyrights. To back this up, the Archive claimed it was using technology “to make lending more convenient and efficient” because its work allowed users to do things that were not possible with physical books, such as permitting “authors writing online articles [to] link directly to” a digital book in the Archive’s library. The Archive also insisted its library was not supplanting the market for the publisher’s products.
The District Court rejected these arguments, holding that no case or legal principle supported the Archive’s defense that “lawfully acquiring a copyrighted print book entitles the recipient to make an unauthorized copy and distribute it in place of the print book, so long as it does not simultaneously lend the print book.” The judge also deemed the concept of Controlled Digital Lending “an invented paradigm that is well outside of copyright law.”
In affirming the District Court’s ruling, the Second Circuit Court applied the four-part test for fair use that looks at: (1) the purpose and character of the use; (2) the nature of the copyright work; (3) the portion of the copyrighted work used (as compared to the entirety of the copyrighted work); and (4) the impact of the allegedly fair use on the potential market for or value of the copyrighted work.
The first factor — the purpose and character of the use — is broken down into two subsidiary questions: Does the new work transform the original, and is it of a commercial nature or is it for educational purposes? Neither the District Court nor the Court of Appeals bought the Internet Archive’s claim that its Open Library was transformative. The Court of Appeals held that the digital books provided by the Internet Archive “serve the same exact purpose as the original; making the authors’ works available to read.” (The Court of Appeals did find that, as a not-for-profit entity, the Internet Archive’s use of the books was not commercial.)
On the second factor, which is generally unimportant here, the Court of Appeals also found in favor of the publishers. Of greater significance is factor three, which looks at how much of the copyrighted work is at issue. Copying a sentence or a paragraph of a book length work is more likely to be fair use than copying the entire book which, of course, is exactly what the Internet Archive was doing. Again, another win for the publishers.
And arguments on factor four — the impact on the market for the publishers’ products — didn’t work out any better for the Internet Archive. Notably, the Court of Appeals found that the Internet Archive was copying the publishers’ books for the exact same purpose as the original works offered by the publisher, thus naturally impacting their market and value.
So what are the takeaways here as we look ahead to the case between the New York Times and Open AI/Microsoft?
On the one hand, OpenAI/Microsoft have copied entire articles from the Times (and the numerous other plaintiffs that are suing OpenAI and Microsoft), which will hurt OpenAI/Microsoft claims of fair use. Likewise, OpenAI/Microsoft’s fair use arguments won’t get very far if the Times can show that ChatGPT’s works are negatively impacting the market for its work or functioning as a substitute for journalism.
On the other hand, if OpenAI/Microsoft can show that ChatGPT’s output transformed the Times’ content, it may be able to prevail on fair use.
In any event, the case between OpenAI and Microsoft and The New York Times is likely to include a lot more ambiguity than in the Internet Archive matter, with the potential to result in new interpretations of copyright law with massive consequences for media and technology companies worldwide.