Powerful new technology continues to surpass human limitations. First, it was listening to the radio. Then, people started watching television. Now, artificial intelligence is being implemented in the workplace. Access to artificially created work comes at the expense of humans being replaced by AI. If the media we consume becomes artificially produced, creatives everywhere are at risk.
On Tuesday, September 19th, The Authors Guild, along with seventeen authors, including John Grisham, George R. R. Martin, and Jodi Picoult, filed a lawsuit against OpenAI for using their copyrighted books to train their large language models (LLMs) that generate text on ChatGPT. The authors did not grant OpenAI a license to use their original work, and they argue that ChatGPT is taking money away from writers. The authors believe ChatGPT would not exist without their writings, and they are seeking financial compensation. However, OpenAI argues that they should be allowed to use the authors’ stories because they serve the transformative purpose of creating technology that surpasses human capabilities.
The Authors Guild was founded over 111 years ago to “protect the rights of all authors.” In 2014, the Guild was awarded an $18-million settlement after the writings of thousands of freelance authors were being resold to electronic databases without the authors’ clearance. While the defendants argued that the revisions made for a transformative work of their own, the court ruled in favor of the Authors Guild.
In 2007 Viacom sued YouTube, arguing that there were not enough measures taken to ensure that the unauthorized clips of Viacom’s copyrighted programs remained off YouTube. The two companies eventually settled outside of court. There is a history of courts protecting existing copyrights, and the Authors Guild believes this case should be no different. The Authors Guild claims that ChatGPT would not exist without the illegal use of their books. They believe authors should have the opportunity to decide if they want their material to be used for generating and teaching artificial intelligence.
OpenAI, founded in 2015 as an artificial intelligence research laboratory, focuses on developing technology to surpass human capabilities. Through models like GPT-4 and DALL-E, OpenAI has advanced artificial image and text generation. In July 2019, Microsoft invested $1 Billion in OpenAI to become their exclusive provider of cloud computing services. Artificial intelligence technology can analyze data, answer questions, and make predictions.
Artificial intelligence is exciting and has a growing field of application. However, it is still in its early stages and has a long way to go. ChatGPT, the primary AI language model of OpenAI, is designed to understand and generate human-like text. ChatGPT has been criticized for displaying a left-leaning bias and a tendency to produce misinformation. The AI service can impersonate authors and write low-quality work that reflects negatively on their reputation. It can be exploited for profit on social media without original authors receiving any compensation. ChatGPT has also been used to create alternate endings, allowing content creators to profit from unoriginal ideas.
There has yet to be a legal precedent for artificial intelligence using copyrighted materials, and perhaps this lawsuit will establish the framework for how to handle copyrights and artificial intelligence.
Copyright Law and Fair Use
The United States first established copyright law in Article 1, Section 8, Clause 8 of the US Constitution, giving the government authority to approve ownership of creative expressions. As digital technologies like television and photography emerged, the Copyright Act of 1976 expanded to include new forms of creative expression.
As the world changed, new norms were put in place to protect the ownership of original work. DVD players stopped allowing users to create copies of discs, video streaming services emerged to purchase exclusive rights to TV shows and movies, cloud storage services have implemented copyright infringement detection, and copyright laws expanded to include blueprints of 3D printed designs.
Fair use is a complex and unpredictable doctrine that is usually assessed on a case-by-case basis. It allows limited use of copyrighted material for a transformative purpose, like critique, parody, and preservation. In the 2019 Google vs. Oracle case, the Supreme Court ruled that Google’s use of Oracle’s Java code in its operating systems was transformative and served a different purpose. As a result, the court determined that Google’s use of Oracle’s code constituted fair use.
If the court determines that OpenAI’s text generation software serves a transformative purpose, fair use would be determined, and the authors would lose the lawsuit. This determining factor will be critical as the lawsuit progresses.
In August of this past year, OpenAI motioned to dismiss two lawsuits by authors Sarah Silverman, Paul Tremblay, Mona Awad, Chris Golden, and Richard Kadrey. They alleged that ChatGPT was illegally trained using pirated copies of their books. OpenAI argued that its model used the authors’ texts as only “a tiny part” of the dataset to teach its models.” OpenAI also claimed that the authors’ lawsuits “misconceive the scope of copyright, failing to take into account the limitations and exceptions (including fair use) that properly leave room for innovations like the large language models (LLMs) now at the forefront of artificial intelligence.”
OpenAI confessed to using the authors’ copyrighted material to train the LLMs that run their artificial intelligence, which would be illegal unless the court ruled in favor of fair use.
In an interview with Rebecca Tushnet, professor of copyright and trademark law at Harvard, she believes that OpenAI has the upper hand in this lawsuit. She thinks artificial intelligence is transformative, and this is a situation where the government should create a new policy to protect existing copyrights while advancing our technological capabilities. Compensation seems plausible because OpenAI is on track to earn over $1 Billion in the next 12 months.
As Artificial Intelligence becomes increasingly used for academic growth and personal entertainment, the era of AI having total knowledge across the internet may be coming to an end. If the court rules that OpenAI can only allow its LLMs to use the information available in the public domain, all websites with AI search engines run the risk of being less useful. The court’s ruling on fair use will be the determining factor.
The lawsuit between The Authors Guild and OpenAI is a pivotal moment on the edge of technology and creativity. The authors’ pursuit of fair compensation is a testament to the preservation and value of creativity. The verdict in this case will mark the beginning of a new era of justice for writers in the digital age.
For more on AI and Writing, try Don’t Feed the Machine: The Truth About AI