Copyright Center

Copyright Implications of AI

While it might sound like a concept straight out of science fiction—think robots and androids—artificial intelligence (AI) is very much a reality that has taken the world by storm. But there's no need to fear a robot apocalypse anytime soon!

Though the technology is still in its early stages, it's advancing rapidly. The recent surge in AI coverage is largely due to startups releasing innovative tools, such as ChatGPT. This AI model responds to prompts with coherent, contextually relevant content and can even recall earlier conversations in the same thread.

To understand how AI generates content, we must delve into a process known as machine learning. AI systems are trained to recognize patterns and skills by analyzing vast databases of publicly available images and text from the internet. This training process involves creating digital copies of existing works. For example, OpenAI states that its programs are trained on "large, publicly available datasets that include copyrighted works," and this involves making copies of data for analysis. As a result, much of this data originates from human-created content, raising various copyright implications.

Creating these copies without explicit permission could infringe upon the copyright holder’s exclusive rights to reproduce or create derivative works. Nevertheless, many AI companies argue that their training methods fall under the copyright doctrine of fair use.

If you’re unfamiliar with fair use, it’s a provision in the U.S. Copyright Statute that allows limited use of copyrighted material without obtaining permission from the copyright owner. The determination of fair use hinges on four statutory factors outlined in 17 U.S.C. § 107:

1. The purpose and character of the use, including whether it is commercial or for nonprofit educational purposes;
2. The nature of the copyrighted work;
3. The amount and substantiality of the portion used in relation to the copyrighted work as a whole; and
4. The effect of the use on the potential market for or value of the copyrighted work.

Many AI companies, including OpenAI, argue that their use is "transformative" rather than merely "expressive," as the training process aims to create a "useful generative AI system." They also maintain that the third factor supports fair use, as the copies are not made public but are solely for training purposes.

However, the most significant contention arises around the fourth factor—impact on market profitability. Critics express concerns that AI can generate works similar to those in the copyrighted datasets it has been trained on. This can lead to competition with original works, potentially undermining their market value. A notable example is the AI-generated song “Heart on My Sleeve,” which mimics the styles of artists Drake and The Weeknd. It garnered millions of streams before Universal Music Group, representing both artists, claimed that AI companies were infringing copyright by using their songs in training data.

For more insights into the potential effects of artificial intelligence on copyright, libraries, and academic publishing, be sure to check out Episode 52 of our podcast, More Than Books, where we discuss AI and AI art.