R.F. Kuang’s Yellowface – Copyright implications within the storyline

A few months ago, I read the book Yellowface by R. F. Kuang. The book is well written, and acts to comment on many issues including diversity and racism, but the main character is painfully unlikeable. In the novel, the main character, Juniper, is an aspiring but average novelist. She is present when her friend, Athena, a successful author, dies in a choking accident. She steals a partially written manuscript before calling for help. She takes the partially written manuscript and develops it, both adding and editing it, until it is transformed into a full-fledged novel based on a huge amount of historical research that she takes on herself. After its publishing, she lives in fear of her acts being discovered. It would harm her reputation, but what of the question of her liability?

            Written works are protected by copyright law. Under s.5(1) of the Copyright Act, original literary works acquire copyright protection. Because Athena’s manuscript was an original literary work in its initial, partially complete stage, it would still attract this type of protection. It would not matter that the manuscript was not edited. Under University of London Press, Ltd. v. University Tutorial Press[1] quality of a work does not determine whether or not it is protected. It is the substance that matters. As such, the fact that Athena’s work was a rough draft, it would still be protected.

            Athena’s unfinished manuscript was based on historical fact relating to the Chinese army. Under Winkler v Hedley, there is no copyright protection over facts. As such, it could be argued that Juniper’s finished product was simply a similar story based over unprotected historical facts, and as such no infringement arises. This was the case in Maltz v. Witterick[2], where it was held that a book based on a real woman’s life did not attract protection from a documentary that covered the same. The question that arises then is how much of the unfinished work was fact and how much was an original work by Athena?

            To attract copyright protection, a degree of originality must be in place.[3] Under CCH Canadian Ltd. v Law Society of Upper Canada[4] to be considered an original work, there should be skill and judgment applied to the exercise. There should be some degree of intellect applied as well. Because Athena’s work was her own story based inside the historical setting, the originality requirement is met, and the work does attract such protection. Summarily, Athena’s work attracts protection on the grounds of qualifying a literary work, and constituting of original storytelling on a foundation of historical fact.

            With that being said, the finished product was not all Athena’s work – it was Juniper’s operating on top of Athena’s. The Act protects against copying in that it gives the copyright owner “the sole right to  the work or any substantial part” of the work.[5] The ruling in Cinar Corp. v. Robinson[6] affirms this statute in stating that infringement is present when a substantial amount of the plaintiff’s material is found in the defendant’s work. In Yellowface, Athena’s ideas formed the structure of the finished novel. But it was not merely her ideas, which would go unprotected by copyright law[7] – her words were the foundation of Juniper’s finished product. While these cases are evaluated on the degree of copying that is present, under Cinar, it is indicated that Juniper’s publishing of the novel would be an act of copyright infringement.

            Yellowface is a fictional story of one author copying the ideas and substance of another. In real life, there are accusations of this sort with relation to AI. Authors are taking court action because companies are taking their work found in certain databases (allegedly illegally accessed), pumping these works into AI programs, and setting the program to spit out similar material[8]. As such, the programs are using the original, copyrighted work and altering and building out on it to produce another work. This is not unlike what Juniper did in Yellowface.

            While the claims of the authors seem to have moral foundation if not a legal one, their actions were dismissed in court.[9] The court stated that because the AI works did not appear to contain a substantial amount of the original copyrighted works, there could be no basis for an infringement claim – this flows from the ruling under Cinar. As such the claim was not allowed to proceed, and it can be inferred that there is no copyright infringement when OpenAI takes an author’s copyrighted work to pump out a similar story. The complexity of this is that there is still something wrong with this morally; intuitively there is something bad about taking someone’s writing and using it for a program to create different, probably slightly less quality writing. But further, some writers are now using AI to write by “incorporating AI-generated text into [their] narrative.”[10] If it is true that copyrighted works are being used to train AI, this means that some authors are using copyrighted material output by a program as scaffolding for their own narrative. While in all likelihood the completed works would fail an infringement action based on the substantiality requirement, there is something harmful about this progression in terms of author credit, and reward hard work – the importance of which is the basis of copyright law.

            Summarily, copyright law is designed protects author’s works, reward authors, and benefit readers by disseminating such works. This short glance at the Yellowface novel outlines the protection offered to unfinished literature that is worked into a larger finished product, and how this protection was not offered to authors who were affected by AI programming in a similar manner to the affect of Juniper’s copying Athena’s work. This suggests a call for greater protection of author’s rights in the sphere of copyright law.

[1] [1916] 2 Ch 601

[2] 2016 FC 524

[3] Copyright Act s.5(1)

[4] [2004] 1 SCR 339

[5] Copyright Act s.3(1)

[6] [2013] 3 SCR 1168

[7] Under Delrina Corp v. Violet Systems Inc. 2002 CanLII 11389 (ON CA), <https://canlii.ca/t/1dnrr>, copyright protection is attracted by expressions of ideas – not simply ideas themselves.

[8] Wes Davis “Sarah Silverman is suing OpenAI and Meta for copyright infringement,” online (2023) The Verge https://www.theverge.com/2023/7/9/23788741/sarah-silverman-openai-meta-chatgpt-llama-copyright-infringement-chatbots-artificial-intelligence-ai

[9] Winston Cho “Sarah Silverman, Authors See Most Claims Against OpenAI Dismissed by Judge,” online (2024) The Hollywood Reporter https://www.hollywoodreporter.com/business/business-news/sarah-silverman-openai-lawsuit-claims-judge-1235823924/

[10] Steven W. Beattie, “What do we want AI to do? Write grant applications. Do our taxes. I don’t want that tool [to help write my books]” (2023) The Toronto Star https://www.thestar.com/entertainment/books/what-do-we-want-ai-to-do-write-grant-applications-do-our-taxes-i-don/article_b83ca983-36d7-5557-8b0f-ddf1eca01ebd.html