By: Miguel E. Serrano

On October 9, 2023, the Writers Guild of America announced that 99% of its members voted to end its historic strike.[1] In May, the Writers Guild and the Screen Actors Guild-American Federation of Television flanked Hollywood and shut down American entertainment.[2] Daytime and late-night talk shows went silent.[3] Canceled press tours heralded the release of Barbie and Oppenheimer.[4] Strikes also paused the development of shows like Stranger Things, The Last of Us, and Abbott Elementary.[5] According to union members, writers sought to prohibit storylines, dialogue, or fully copyrighted works from becoming “source material” in A.I. training data and to stop A.I. writing-software from replacing on-the-ground writers.[6]

The Writers Guild’s victory is a chapter in American labor history, a workers’ declaration of rights when the law on A.I. was still unwritten. Creators allege that an A.I.’s ability to produce faithful and precise summaries of their work is proof that copyrighted content was “mined” to train models without permission, infringing on an author’s copyright.[7] Creators also argue that A.I. output are unlawful derivative works that infringe on the original author’s copyright.[8]After 146-days of striking, and an estimated $5 billion dollars in lost revenue,[9] executives were forced to the negotiating table, agreeing to a three-year deal with the Writers Guild placing strict limits on A.I.-written scripts.[10]  Now, the Authors Guild has published an open letter signed by over 8,500 prominent authors – including Margaret Atwood, George R.R. Martin, and Dan Brown – requesting that generative A.I. products cease to use their works without proper licensing and requisite compensation.[11]

Although case law is developing, in a hearing before a California District Court, the court was initially skeptical about the broad-ranging arguments raised by creators suing generative A.I. companies, stating that plaintiffs did not differentiate infringement claims against different defendants, and that plaintiffs required more facts to articulate copyright infringement.[12] According to the hearing, the court also disagreed that copyrighted works and the A.I. outputs were substantially similar.[13]

Companies have also asserted a notable exception to copyright infringement: fair use.[14] In response to lawsuits, companies argue that Large Language Models “generate original expression [that] is transformative by nature and quintessential fair use,” which serves public policy by allowing novel innovations like A.I. to develop.[15]

Is it true that OpenAI’s GPT-4, Google’s Bard, and Meta’s Llama 2 are “partly built on millions of books, articles, online chats[,] and other content posted online.”[16] But these Large Language Models (LLMs) train on public, non-copyrighted content, such as by mass scraping articles on Wikipedia and conversations on Reddit.[17] However, LLMs rely on high-quality, human-generated content to train and perform well – to predict the likelihood that a phrase, sentence, or paragraph will properly answer a prompt.[18] A response to a prompt is generally novel, meaning it is not a carbon copy any individual piece of content from the training data.[19] But other times the output can have an uncanny degree of likeness to training data.[20] Outlets like The New York Times and Twitter have responded to this potential encroachment by affirmatively blocking web crawlers.[21]

In an August announcement, the U.S. Copyright Office opened up a public comment period to inform its ongoing study of A.I. tools.[22] The Office held a series of listening sessions earlier this year with representatives of A.I. companies, such as Microsoft, a major investor in OpenAI, and members of The Authors Guild.[23] The Office is pursuing comments until October 18, potentially before issuing regulatory action or new federal rules to address the intersection of intellectual property and A.I.: acquisition of training data, whether A.I. generated material is eligible for copyright, liability for infringement by A.I. systems, and how to treat A.I. outputs that imitate human artists.[24]

The popularity of automated computing and information technology has excited the collective imagination.[25]There is an industrial-like splendor to its speed, and shock when this young technology delivers mature and quality art.[26]  Accordingly, the law on A.I. may develop under an audacious third path: SILO, a new language model that attributes particular results to authors.[27] Developed by researchers at the University of Washington in Seattle, UC Berkeley, and the Allen Institute for AI, authors can remove content from the A.I.’s training set and license limited access to their intellectual property.[28] SILO attempts to mitigate legal risks by properly crediting and at times paying owners when their data is used.[29]

[1] Hollywood writers vote to approve contract deal that ended strike as actors negotiate, AP News (Oct. 9, 2023), https://apnews.com/article/writers-strike-hollywood-contract-actors-negotiations-43a57ce4783a5615c359db1091e0fa89.

[2] Andrew Dalton, As writers reach tentative deal and actors look ahead, here are the Hollywood Strikes’ key players, AP News (Sep. 25, 2023), https://apnews.com/article/hollywood-strikes-lombardini-drescher-stutzman-iger-zaslav-00c17264129cebb17d10b80b9c51aa3e.

[3] Laurence Peter and Steven McIntosh, Hollywood writers in deal to end US studio strike, BBC News (Sep. 25, 2023), https://www.bbc.com/news/world-us-canada-66909250.

[4] Id.

[5] Id.

[6] See Andrew Dalton, AI is the wild card in Hollywood’s strikes. Here’s an explanation of its unsettling role, AP News (July 21, 2023), https://apnews.com/article/artificial-intelligence-hollywood-strikes-explained-writers-actors-e872bd63ab52c3ea9f7d6e825240a202.

[7] See Blake Brittain, Lawsuit says OpenAI violated US authors’ copyrights to train AI chatbot, Reuters (June 29, 2023), https://www.reuters.com/legal/lawsuit-says-openai-violated-us-authors-copyrights-train-ai-chatbot-2023-06-29/.

[8] See e.g., Plaintiffs’ Complaint at 17–35, Authors Guild et al. v. OpenAI Inc. et al., No. 1:23-cv- 08292, (S.D. Cal. Sept. 19, 2023).

[9] Laurence Peter and Steven McIntosh, Hollywood writers in deal to end US studio strike, BBC News (Sep. 25, 2023), https://www.bbc.com/news/world-us-canada-66909250.

[10] Ryan Ozawa, Writers Guild, Studios Reach Tentative Deal Covering AI, Streaming Services, Decrypt (Sep. 24, 2023), https://decrypt.co/198561/writers-guild-strike-tentative-deal-ai-streaming-amptp.

[11] Ben Lutkevich, AI lawsuits explained: Who’s getting sued?, TechTarget (Aug. 4, 2023), https://www.techtarget.com/whatis/feature/AI-lawsuits-explained-Whos-getting-sued.

[12] See generally H’rg on Motion to Dismiss, Andersen et al v. Stability AI Ltd. et al., No. 3:23-cv-00201 (N.D. Cal. July 9, 2023).

[13] See id; see also Isaiah Poritz, AI Art Generators Likely to Notch Early Win in Copyright Case, Bloomberg Law(July 19, 2023), https://www.bloomberglaw.com/bloomberglawnews/tech-and-telecom-law/X6J76EK0000000?.

[14] See Defendant’s Motion to Dismiss at 2, Kadrey et al. v. Meta Platforms, Inc., No. 3:23-cv-03417, 2023 U.S. Dist. Ct. Motions Lexis 229596 (N.D. Cal. July 7, 2023) (citing Authors Guild v. Google, Inc., 804 F.3d 202 (2d Cir. 2015)). In 2015, the Second Circuit ruled in that Google’s digitized copies of millions of books and public snippets of their text was protected under fair use. While the Authors Guild sued Google claiming that the storage and public snippets of authors’ works was copyright infringement, the court held that Google’s software was transformative and did not provide a market substitute for the books in question.

[15] Blake Brittain, AI companies ask U.S. court to dismiss artists’ copyright lawsuit, Reuters (Apr. 19, 2023), https://www.reuters.com/legal/ai-companies-ask-us-court-dismiss-artists-copyright-lawsuit-2023-04-19.

[16] Alistair Barr, ‘Data leverage’ and the Harry Potter test: How much is a single book worth to a giant AI model?Business Insider (Aug. 30, 2023), https://www.businessinsider.com/data-leverage-harry-potter-test-putting-value-data-ai-models-2023-8.

[17] Supra Lutkevich, note 11.

[18] Adam Jermyn, et. al., Conditioning Predictive Models: Large language models as predictors, Less Wrong (Feb. 2, 2023), https://www.lesswrong.com/posts/XwXmedJAo5m4r29eu/conditioning-predictive-models-large-language-models-as.

[19] Rahul Matthan, Does AI Copy?, Ex Machina (Feb. 15, 2023), https://exmachina.substack.com/p/does-ai-copy.

[20] See e.g., Plaintiff’s Complaint at 18, Getty Images, Inc. v. Stability AI, Inc., No. 1:99-mc-09999 (D. Del. Feb. 3, 2023) (“[T]he output delivered by Stability AI includes a modified version of a Getty Images watermark, underscoring the clear link between the copyrighted images that Stability AI copied without permission and the output its model delivers.”).

[21] Kali Hays, The US Copyright Office just took a big step toward new rules for generative AI, Business Insider(Aug. 30, 2022), https://www.businessinsider.com/us-copyright-office-new-rules-generative-ai-2023-8.

[22] A Notice by the Copyright Office, Library of Congress, Federal Register (Aug. 30, 2023), https://www.federalregister.gov/documents/2023/08/30/2023-18624/artificial-intelligence-and-copyright.

[23] Supra Hays, note 21.

[24] Supra Brittain, note 15.

[25] Hillel Italie, Fiction writers fear the rise of AI, but also see it as a story to tell, AP News (Aug. 13, 2023) (“I’m frightened by artificial intelligence, but also fascinated by it. There’s a hope for divine understanding, for the accumulation of all knowledge[.]”), https://apnews.com/article/ai-novels-authors-f846ff75df37dacbdb87b4343a346c2a.

[26] Vauhini Vara, Confessions of a Viral AI Writer, Wired (Sep. 21, 2023) (“I’d often heard the argument that AI could never write quite like a human precisely because it was a disembodied machine. And yet, here was as nuanced and profound a reference to embodiment as I’d ever read. Artificial intelligence had succeeded in moving me with a sentence about the most important experience of my life.”), https://www.wired.com/story/confessions-viral-ai-writer-chatgpt/.

[27] Supra Hays, note 21.

[28] Tero Ojanpera, Timo Vuori, and Quy Nguyen Huy, A.I. researchers: ‘If we want A.I. to be good, we must nurture it like we would a child’, Fortune (June 14, 2023), https://fortune.com/2023/06/14/ai-good-nurture-it-like-we-would-a-child-tech/.

[29] Id.