The use of humanity using copyrighted books in the AI training process was “very transformative” and used fairly, U.S. Senior District Judge William Alsp ruled Monday. This is the first time a judge has given AI companies a big victory for a generative AI company and has hit creators with a big blow.
Fair use is a doctrine that is part of US copyright law. This is a four-part test that allows people and businesses to use protected content for specific purposes without the permission of the rights holder, such as when the criteria are met. High-tech companies say that fair use exceptions are essential to access the large amount of human-generated content needed to develop the most sophisticated AI systems.
Writers, actors, and many other types of creators were equally clear in arguing that the use of content to promote AI is not a fair use. Publishers, artists and content catalog owners have filed lawsuits alleging that AI companies such as Openai, Meta and Midjourney are infringing protected intellectual property in order to avoid expensive but standard licensing procedures. (Disclosure: CNET's parent company Ziff Davis filed a lawsuit against Openai in April, claiming it infringed Ziff Davis' copyright in training and operating AI systems.)
The author, who is suing humanity for copyright infringement, says they also obtained their books illegally. In other words, the book was pirated. It leads to the second part of Alsup's ruling, based on his concerns about how to obtain books on humanity. In the ruling, he wrote that Ben Mann, co-founder of humanity, intentionally downloaded illicit copies of five million books from Libgen and another two million copies from Pirate Library Mirror (Pirlimi).
The ruling also outlines how humanity intentionally obtained printed copies of previously pirated books to create “their own catalogue of bibliographic metadata.” The verdict says that Tom Turbay, vice president of humanity, “we are avoiding “legal/practical/business throws” while getting all the books in the world.” The human team destroyed millions of used books in this process, discarded them, prepared for machine-readable scans, stripped them from the bindings and chopped them down to fit.
The acquisition and digitalization of human print books was a fair use, the verdict says. However, “creating permanent, generic libraries in itself was not a fair use that allowed humanity to infringe copyright infringement.” Alsup ordered a new trial on pirated libraries.
Humanity is one of many AI companies facing copyright claims in court, so this week's ruling could have a major ripple effect across the industry. Before you know how your copyright infringement claims are ordered to pay damages, you need to see how your copyright infringement claims will be resolved. However, if scales are leaning towards giving fair use exceptions for multiple AI companies, the creative industry and those working within it will certainly suffer.
For more information, see our guide to Understanding Copyrights in the Age of AI.
