by Elias Aidun, Associate Member, University of Cincinnati Law Review Vol. 93
I. Introduction
Copyright is a type of intellectual property that has been a part of U.S. law since the nation’s founding.[1] The primary purpose of copyright law is to promote the creation and distribution of works for the benefit of the public.[2] “To promote the progress of science and useful arts” the government extends copyright protection to authors’ original works.[3] Specifically, copyrights are designed to protect original works of authorship, such as paintings, photographs, illustrations, musical compositions, sound recordings, computer programs, books, poems, blog posts, movies, architectural works, plays, and much more.[4] By granting authors the exclusive right to authorize certain uses of their works, copyright provides economic incentives to those authors to create new works and make them available in the marketplace.[5]
A copyright for an original work provides the author with ownership in the exclusive right to reproduce the work, prepare derivative works, distribute copies of the work to the public by sale or other transfer of ownership, perform the work publicly, and display the work publicly.[6] Although copyright owners are provided these exclusive rights, the legal doctrine of fair use may permit unlicensed use of copyright-protected works.[7] Under 17 U.S.C § 107, fair use is an affirmative defense to copyright infringement and permits a potentially infringing party to use copyrighted work without the copyright owner’s permission in certain circumstances.[8]
A notorious example of fair use of a copyrighted work comes from Sony’s Video Cassette Recorder (“VCR”).[9] VCRs allowed consumers to record television and movies, including copyrighted works, but film companies eventually sued Sony for copyright infringement.[10] The Supreme Court, in a 5-4 decision, ultimately held that Sony’s VCRs were a fair use of copyrighted works.[11] While Sony’s VCR survived by a successful fair use claim 40 years ago, OpenAI and it’s large-language models’ (“LLMs”) fate is currently up in the air.[12] In December 2023, The New York Times (“The Times”) filed a copyright lawsuit against OpenAI and its partner, Microsoft, for copying and using millions of The Times’ copyrighted works to build the knowledge base for their LLM, ChatGPT.[13]
This article will explore the potential impact that The New York Times Company v. Microsoft Corporation will have on the future of LLMs and its regulation under copyright law. Part II provides background on copyright law and The Times’ complaint. Part III discusses court precedent, regarding fair use arguments, and how it could be applied to The New York Times Company. Finally, Part IV offers a brief conclusion on how The New York Times Company could set the precedent for future LLMs and how they are regulated under copyright laws.
II. Background
As artificial intelligence (“AI”) continues to rapidly improve, it is astonishing how much knowledge AI models contain and how thoughtful their responses are. For example, ChatGPT can break down and explain complex topics such as computer programming or dark matter into layperson terms, or provide the latest news updates around the world.[14] Regardless of the question asked, ChatGPT most likely has a cohesive response. But how exactly does ChatGPT know so much? Well, according to The Times, ChatGPT’s knowledge base was created by using The Times’, and other content creators’ copyrighted works.[15] Within the suit, The Times alleges that ChatGPT, when prompted by users, can generate responses that contain “near-verbatim copies” of “significant portions” of The Times’ copyrighted material.[16] According to The Times, all this violates copyright law and undercuts The Times’ business model, which relies on subscriptions, advertisement revenue, and licensing.[17]
A. Copyright Law
As mentioned above, copyright is a type of intellectual property that protects an author’s original work.[18] For a work to qualify for copyright protection, it must satisfy two requirements under 17 U.S.C § 102(a): originality and fixed in a tangible form of expression.[19]
Original works of authorship must be created by a human author and have a minimal degree of creativity.[20] Famously stated in Feist Publ’ns, Inc. v. Rural Tel. Serv. Co., an original work must have a “spark” and “modicum” of creativity.[21] Therefore, the originality requirement is understood to be a low threshold to satisfy. Nevertheless, some works are simply not creative, and thus not eligible for copyright protection. Within 17 U.S.C § 102(a), Congress specifically stated that copyright protection does not extend to any “idea, procedure, process, system, method of operation, concept, principle or discovery.”[22] Rather, copyrights are designed to protect the author’s expression.
Works of authorship are fixed in a tangible form of expression when the work is captured in a sufficiently permanent medium so that the work can be perceived, reproduced, or communicated for more than a short time.[23] As it relates to The Times, journalists become copyright owners once their work is written down or recorded.[24]
B. Fair Use of Copyrights
Within copyright law, the legal doctrine of fair use is designed to promote the freedom of expression by permitting the unlicensed use of copyright-protected works in certain circumstances.[25] 17 U.S.C § 107 provides the legal framework for fair use and identifies certain types of use that may qualify for fair use – such as criticism, comment, news reporting, teaching, scholarship, and research.[26] To determine whether a use of a copyrighted work is a fair use, 17 U.S.C § 107 provides four factors to consider: 1) the purpose and character of the use, including whether such use is of a commercial nature or is 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 upon the potential market for or value of the copyrighted work.[27] Within fair use jurisprudence, a close relation exists between the first factor (purpose and character of the use) and the fourth factor (market effects).[28] Additionally, courts have described the market effects factor to be “undoubtedly the single most important element of fair use.”[29]
C. The New York Times Complaint
On December 27, 2023, The Times filed a copyright infringement lawsuit against Microsoft and its partner OpenAI for “unlawful use of The Times’ work to create artificial intelligence products,” specifically LLMs.[30] Within the complaint, it alleges that Microsoft and OpenAI’s LLMs, such as ChatGPT, “were built by copying and using millions of The Times’ copyrighted news articles, in-depth investigations, opinion pieces, reviews, how-to guides, and more.”[31] Although Microsoft and OpenAI copied from many sources to build their LLMs, The Times claims that Microsoft and OpenAI gave particular emphasis to The Times’ content demonstrating the value of The Times’ works.
Specifically, The Times claims that Microsoft and OpenAI’s LLMs “can generate outputs that recite Times content verbatim, closely summarize it, mimic its expressive style,” and even wrongly attribute false information to The Times.[32] To emphasize this point, The Times included an example from a 2012 article.[33] The 2012 Times article, in an enormous effort across three continents, described how Apple and other technology companies transformed the global economy through outsourcing.[34] OpenAI had no role in the creation of this article, nonetheless, with minimal prompting, OpenAI’s LLM copied large portions of the article verbatim.[35] In addition, OpenAI’s LLM directly copied paragraphs of Times articles that were only available to readers through a paywall.[36]
According to The Times, the unlicensed use of its content by OpenAI and its LLMs competes with and threatens The Times’ ability to provide trustworthy information, news analysis, and commentary.[37] More importantly, The Times claims that OpenAI and Microsoft “seek to free-ride on The Times’ massive investment in its journalism” by building substitute products without permission or payment.[38]
III. Discussion
The Times is not alone in the fight against ChatGPT. Other well-known authors like George R.R. Martin, John Grisham, and Jodi Picoult are also suing ChatGPT in a class action for “systematic theft on a mass scale.”[39] As The Times and authors’ arguments rely on copyright infringement, OpenAI’s response rests on the legal doctrine of fair use. However, in previous cases, courts have largely sided with technology companies in interpreting how copyright laws should treat AI models.[40]
A. Modern Copyright Precedent
In 2023, three visual artists brought suit against Stability AI, a London-based maker of text-to-image generators, alleging that AI image generators violate the rights of millions of artists by ingesting digital images and producing derivative works that compete against the originals.[41] However, in a defeat for the visual artists, a federal judge in San Francisco dismissed much of the lawsuit.[42] Similarly, another California judge dismissed part of comedian Sarah Silverman’s copyright claims against Meta.[43] To combat these copyright claims, AI and tech companies have relied on Google’s past success.[44]
In The Authors Guild v. Google, Inc., an association of authors of copyrighted books sued Google for making digital copies of tens of millions of books without permission.[45] Google made digital copies of books, that were not previously online, to launch its Google Books project.[46] Internet users could go to the Google Books website and search for terms within the books.[47] Users could also pull up a limited viewing of small portions of the book’s text.[48] Ultimately, the court held that Google’s unauthorized digitizing of copyright-protected works, creation of a search functionality, and display of snippets from those works are non-infringing fair uses.[49] The court reasoned that the purpose of the copying is highly transformative, the public display of text is limited, and the revelations do not provide a significant market substitute for the protected aspects of the originals.[50]
B. The New York Times v. OpenAI Fair Use Analysis
The Times’ arguments, taken in isolation, might be highly convincing that copyright infringement occurred; however, OpenAI’s fair use defense might invalidate any liability.[51] Specifically, this case might come down to two factors of the fair use analysis; purpose and character of the use, and market effects.
When analyzing the purpose and character of the secondary use, the inquiry should focus on whether the new work merely supersedes the objects of the original creation, or instead adds something new, with a further purpose.[52] In other words, to what extent is the new work transformative.[53] In support of OpenAI’s defense, ChatGPT has numerous other purposes besides copying The Times’ exclusive articles. ChatGPT can be used in professional careers ranging from education to marketing and even healthcare.[54] Specifically within education, ChatGPT can help educators by creating instructional content, offering suggestions, and acting as an online educator to learners by answering questions.[55] Especially with The Authors Guild in the backdrop, OpenAI could have a valid defense that ChatGPT has a transformative purpose. However, ChatGPT’s ability to reproduce large, verbatim excerpts of The Times’ works to bypass a paywall, will most likely not be considered a transformative use.[56]
Even if the first factor (purpose and character of the use) of the analysis does not favor a fair use, the fourth factor (market effects) is “undoubtedly the single most important element of fair use.”[57] The determination of market effects focuses on actual or potential market substitution.[58] The Times’ argument here is if OpenAI did not copy millions of The Times’ copyrighted works, to build the knowledge base of ChatGPT, then OpenAI would have to pay The Times license fees for the articles.[59] Although avoiding license fees is not necessarily a market harm, The Times argues that ChatGPT’s ability to reproduce The Times’ articles verbatim and bypass paywalls displaces the market and creates a substitution.[60] Thus, OpenAI and ChatGPT are directly depriving The Times of revenue.[61] Nevertheless, it still might be difficult for The Times to demonstrate that ChatGPT is a meaningful substitute for the historic publication. If an individual wanted to read an article from The Times, would they really go through the effort to ask ChatGPT-specific questions to reproduce the article on a paragraph-by-paragraph basis?[62] Or would the individual simply pay The Times for access to the article?[63] These inquiries make this case ripe for debate.[64]
IV. Conclusion
As AI technology advances, traditional intellectual property laws struggle to address issues of authorship, ownership, and the nature of creativity.[65] The New York Times Company case could have the potential to set the precedent for future AI technology, specifically LLMs, and how they are regulated under copyright laws. However, because of modern copyright precedent and copyright law, The Times will have a difficult challenge in proving that ChatGPT is an unfair use of copyrighted material. Nevertheless, the future of intellectual property law and AI involves grasping the evolving legal landscape and its implications; prompting legal professionals, businesses, and creators to embrace the dynamic and ever-changing legal landscape.[66]
[1] U.S. Const. art 1, § 8, cl. 8.
[2] Copyright Basics, U.S. Pat. and Trademark Office, https://www.uspto.gov/ip-policy/copyright-policy/copyright-basics (last visited Oct. 7, 2024).
[3] Id.
[4] What is Copyright?, U.S. Copyright Office, https://www.copyright.gov/what-is-copyright/ (last visited Oct. 7, 2024).
[5] Copyright Basics, supra note 2.
[6] What is Copyright?, supra note 4.
[7] U.S. Copyright Office Fair Use Index, U.S. Copyright Office, https://www.copyright.gov/fair-use/ (last visited Oct. 7, 2024).
[8] 17 U.S.C § 107.
[9] Sony Corp. of Am. v. Universal City Studios, Inc., 464 U.S. 417 (1984).
[10] Id.
[11] Id.
[12] Id., Complaint, N.Y. Times v. Microsoft Corp., No. 1:23-cv-11195 (S.D.N.Y filed Dec. 27, 2023).
[13] Complaint, supra note 12.
[14] Maxwell Timothy, 15 Things You Can Do With ChatGPT, Make Use Of, https://www.makeuseof.com/things-you-can-do-with-chatgpt/ (Aug. 18, 2024).
[15] Complaint, supra note 12.
[16] Id at 98.
[17] Rachel Reed, ChatNYT, Harvard Law Today (May 22, 2024), https://hls.harvard.edu/today/does-chatgpt-violate-new-york-times-copyrights/.
[18] Copyright Basics, supra note 2.
[19] 17 U.S.C. § 102(a).
[20] What is Copyright?, supra note 4.
[21] Feist Publ’ns, Inc. v. Rural Tel. Serv. Co., Inc., 499 U.S. 340, 362 (1991).
[22] 17 U.S.C. § 102(a).
[23] What is Copyright?, supra note 4.
[24] Id.
[25] U.S. Copyright Office Fair Use Index, supra note 7.
[26] 17 U.S.C § 107.
[27] Id.
[28] Zachary H. Valentine, New York Times vs. OpenAI: Fair Use Fight with Billions at Stake, DarrowEverett (Jan 22, 2024), https://darroweverett.com/new-york-times-vs-open-ai-fair-use-legal-analysis/.
[29] Harper & Row, Publishers, Inc. v. Nation Enter., 471 U.S. 539, 566 (1985).
[30] Complaint, supra note 12, at 2.
[31] Id.
[32] Id at 4.
[33] Id. at 31.
[34] Id.
[35] Id.
[36] Id. at 33.
[37] Id. at 2.
[38] Id. at 2.
[39] Hillel Italie, ‘Game of Thrones’ creator and other authors sue ChatGPT maker OpenAI for copyright infringement, Associated Press, https://apnews.com/article/openai-lawsuit-authors-grisham-george-rr-martin-37f9073ab67ab25b7e6b2975b2a63bfe (Sep. 21, 2023, 2:03 PM).
[40] Matt O’Brien, ChatGPT-maker braces for fight with New York Times and authors on ‘fair use’ of copyrighted works, Associated Press, https://apnews.com/article/openai-new-york-times-chatgpt-lawsuit-grisham-nyt-69f78c404ace42c0070fdfb9dd4caeb7 (Jan. 10, 2024, 4:05 AM).
[41] Jocelyn Noveck & Matt O’Brien, Visual artists fight back against AI companies for repurposing their work, Associated Press, https://apnews.com/article/artists-ai-image-generators-stable-diffusion-midjourney-7ebcb6e6ddca3f165a3065c70ce85904 (Aug. 31, 2023, 2:55 PM).
[42] Matt O’Brien, supra note 40.
[43] Id.
[44] Id.
[45] Authors Guild v. Google, Inc., 804 F.3d 202, 207 (2d Cir. 2015).
[46] Id.
[47] Id.
[48] Id.
[49] Id. at 229.
[50] Id. at 229.
[51] New York Times vs. OpenAI: Fair Use Fight with Billions at Stake, supra note 28.
[52] Authors Guild, 804 F.3d at 214.
[53] Id.
[54] Carolyn Crist, ChatGPT usage at work has more than doubled in a year, HR Dive (Jan. 31 2024), https://www.hrdive.com/news/chatgpt-usage-at-work-doubled/706127/#:~:text=Marketing%20had%20the%20highest%20adoption,67%25%20of%20those%20in%20advertising.
[55] Sukhpal Singh Gill et al., Transformative effects of ChatGPT on modern education: Emerging Era of AI Chatbots, 4 Internet of Things and Cyber-Physical Systems 19, 20 (2024).
[56] New York Times vs. OpenAI: Fair Use Fight with Billions at Stake, supra note 28.
[57] Harper & Row, Publishers, Inc. v. Nation Enter., 471 U.S. 539, 566 (1985).
[58] Id.
[59] Reed, supra 17.
[60] Id.
[61] AI Lawsuit May Eclipse Claims of Fair Use, HodgsonRuss (Apr. 5, 2024), https://www.hodgsonruss.com/newsroom/publications/ai-lawsuit-may-eclipse-claims-of-fair-use.
[62] New York Times vs. OpenAI: Fair Use Fight with Billions at Stake, supra note 28.
[63] Id.
[64] Id.
[65] Sarah Hess, The Evovling Landscape of IP Law in the Age of AI, ABC Legal (Aug. 19, 2024), https://www.abclegal.com/blog/the-evolving-landscape-of-ip-law-in-the-age-of-ai.
[66] Id.
Cover Photo by Steve Johnson on Unsplash.
