AI in Developing Copyright Law

[1]

Transformation from Physical Books to Digital Platforms

Artificial Intelligence (“AI”) is a computer system used for algorithms, search engines, and general assistance, among other applications.[2] Al platforms have become widely prevalent, and this increased use and access to AI technology has caused conflicts between AI creators and human creators.[3] In a suit filed in 2023, Richard Kadrey, an author, comedian Sarah Silverman, and others sued Meta Platforms, Inc. for allegedly unlicensed “training” of Meta’s large language model.^[4] “Training AI” refers to the process of feeding specific AI algorithms information to enhance the system and increase its ability to give users the most accurate results.[5] Meta’s main defense was fair use, arguing that AI creators should be able to use texts from books, song lyrics, or jokes from published comedians without permission while creating algorithms.[6]

Fair Use

The Fair Use exception allows individuals, under limited circumstances, to use copyrighted works without a license.[7] A fair use analysis is composed of four factors “(1) the purpose and character of the use, (2) the nature of the work copied, (3) the amount of the work copied, and (4) the effect of the use on the potential market for the work.”[8] While factor two clearly favors the Plaintiffs, as their work was “expressive” and the type of work protected by the Copyright Act, the court in Kadrey v. Meta Platforms, Inc. ultimately ruled in Meta’s favor.[9] The court’s ruling was focused on whether Meta’s use was transformative, meaning it added a new purpose, expression, or message.[10]

Here, Meta’s use of the works was to train AI models capable of generating diverse text and performing a wide range of functions, rather than to reproduce or distribute the plaintiff’s work.[11] Because the purpose differed so substantially, the court found the training use transformative even though Meta’s system ingested entire works.[12] Looking at market effects, the plaintiffs argued Meta’s conduct harmed the potential licensing market for their works.[13] The court rejected that argument as speculative, noting that the plaintiff had not shown concrete evidence of market harm.[14]

The judge suggested that a stronger argument would have been that AI systems trained on copyrighted material could eventually dilute the market for original works enabling others to generate derivative or stylistically similar content without compensation.[15] Essentially, future artists will be able to create work more easily and efficiently with AI by indirectly using other artists’ work through the algorithms, which is all done without the artists’ consent.[16] AI will indirectly compete with real artists by using the artist's work.[17]

Despite ruling in favor of Meta, the judge noted that Meta’s use of copyrighted works was not technically lawful, but the Plaintiffs’ arguments were not strong enough to favor their position.[18] As a result, there has been deviation from this holding.[19]

Crediting Authors

In another key case, Bartz v. Anthropic PBC, authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson sued Anthropic for copyright infringement.[20] This case, which ended in a settlement, demonstrated the tedious and prevalent nature of lawsuits involving infringement and AI in the future.[21] As per the proposed settlement, Anthropic AI agreed to pay $1.5 billion to settle a copyright infringement suit.[22] If the court approves the settlement, authors will receive about $3,000 for each of the books used in Anthropic AI’s system (about 500,000 books).[23] The Fair Use doctrine is used again in this case, but the authors are starting to fight against it.[24] Ultimately, the hinge point in this case was Anthropic’s use of seven million pirated books without author consent or compensating the authors.[25] However, generative AI copyright infringement cases are more complicated when the work involves comedians, musical artists, and prominent researchers as seen in the Meta case described above.[26]

Looking Toward the Future

Since text is needed to train AI to give responses, the Courts still need to determine where to draw the line between fair use and infringement.[27] AI creators should be able to pull from other works to develop the best possible resource for the public.[28] However, AI creators are also subjected to limitations, as seen above, the fair use doctrine will not protect against dilution.[29] As for the degree of use, that is a topic for the future courts to decide.[30] The area of law is evolving in tandem with technological advancements.[31] In the end, an efficient way to collect the appropriate amount of data with the original creator’s permission is crucial.[32] 

References

[1] Photo by Igor Omilaev, A Computer Chip with the Letter A on Top of It, Unsplash (Oct. 17, 2023),  https://unsplash.com/photos/a-computer-chip-with-the-letter-a-on-top-of-it-eGGFZ5X2LnA.

[2] See NASA, What is Artificial Intelligence?, NASA (May 13, 2024), https://www.nasa.gov/what-is-artificial-intelligence/.

[3] See Chloe Veltman, Anthropic Settles with Authors in First-of-its-Kind AI Copyright Infringement Lawsuit, NPR (Sep. 5, 2025, 8:19 PM), https://www.npr.org/2025/09/05/nx-s1-5529404/anthropic-settlement-authors-copyright-ai.

^[4] See AI Infringement Case Updates: July 7, 2025, McKool Smith (July 7, 2025), https://www.mckoolsmith.com/newsroom-ailitigation-30.

[5] See Michael Chen, What Is AI Model Training & Why Is It Important?, Oracle (Dec. 6, 2023), https://www.oracle.com/artificial-intelligence/ai-model-training/.

[6] See Veltman, supra note 3.

[7] See 17 U.S.C. §107 (2018).

[8] See 17 U.S.C. §107(1)-(4); see also AI Infringement Case Updates: July 7, 2025, supra note 4.

[9] See AI Infringement Case Updates: July 7, 2025, supra note 4.

[10] See id.

[11] See Kadrey, et al. v. Meta, 788 F.Supp.3d 1026, 1044 (2025).

[12] See AI Infringement Case Updates: July 7, 2025, supra note 4.

[13] See id.

[14] See id.

[15] See id.

[16] See id.

[17] See id.

[18] See id.

[19] See id; See also Veltman, supra note 3.

[20] See Veltman, supra note 3.

[21] See id.

[22] See id.

[23] See id.

[24] See id.

[25] See id.

[26] See id.

[27] See Veltman, supra note 3.

[28] See Chen, supra note 5.

[29] See AI Infringement Case Updates: July 7, 2025, supra note 4.

[30] See Hector Craigson, Navigating the Future: Key Insights from the Latest AI Lawsuit Trends, Tech Announcer (May 27, 2025), https://techannouncer.com/navigating-the-future-key-insights-from-the-latest-ai-lawsuit-trends/#:~:text=In%202024%2C%20we%20saw%20a%20surge%20in%20lawsuits,future%20of%20legal%20challenges%20in%20this%20evolving%20field.

[31] See id.

[32] See id.