The AI Millennium: How the DMCA Handles Artificial Intelligence’s Myriad Copyright Issues

Jackson Weist, Contributing Member 2024-2025

Intellectual Property and Computer Law Journal

I. Introduction

In February 2025, the United Kingdom proposed laws that would relax their copyright law scheme by allowing free use of copyrighted materials without permission when used for training artificial intelligence (AI) models.[1] The proposed laws prompted stark backlash from several high-profile British artists like Elton John and Dua Lipa.[2] The policymakers behind the UK laws claimed that the current landscape of UK copyright law was holding back both the development of creative industries and the development of AI.[3] Those who oppose the proposed changes claim that the new laws would hand the life’s work of the entire country’s artists over to AI companies for free.[4] This debate in the UK sets the scene for the current tension between copyright law and technological advancement. Tension between the interests of copyright holders and emerging technologies has been in the backdrop of the copyright world for decades, with a similar turning point taking place during the rise of the internet in the late 1990s.[5] In 1998, the United States passed the Digital Millennium Copyright Act (DMCA), which sought to adapt the US copyright regime to the emerging world of the internet.[6] Now, nearly 30 years later, artificial intelligence poses a similar challenge to lawmakers: is the DMCA enough to account for this new world of AI?

This article analyzes whether the DMCA is equipped to handle the rapidly developing field of artificial intelligence. Part II provides background on copyright law, the DMCA, and how both help and hinder creativity and innovation. Part III discusses how the DMCA may impact the current landscape of artificial intelligence, both in the development of AI models and the content they generate. Part IV concludes by determining if the DMCA does enough to help balance the interests of copyright holders against the interests of AI companies.

II. Background

Copyright Law

In the United States, copyright law provides the creators of artistic works with legal protection over their creations.[7] Creators are given several exclusive rights and control over: reproduction of the work, derivative works, distribution of copies of the work, the work’s public performance and display, and in the case of sound recordings, the digital transmission of the work.[8] For example, an artist who makes a wholly original song has complete control over reproductions and public performances of that song.[9] Copyright as a field of law is founded on the policy view that we value the arts for what they provide to society and want to incentivize creativity without fear of copying.[10] Creatives benefit from extra protection for their works, while society benefits from an enriched creative culture and a rich public domain once copyright protection runs out.[11]

However, some believe the current copyright structure in the US can also hinder creativity and innovation by restricting access to works.[12] Copyright stifles creativity and expressive works in several ways.[13] The mere threat of a copyright infringement suit is often enough of a deterrent for artists and inventors to self-censor and avoid or abandon projects before the law comes to them.[14] There are also several ways that copyright is an attractive weapon to chill speech and silence criticism: the ease of finding infringement, the simplicity of asserting a prima facie infringement case, the uncertainty of defenses like fair use, and the threat of heavy penalties if found guilty.[15] The biggest defense to copyright infringement, fair use, is applied by courts in often subjective and difficult to predict manners, making many would-be infringers hesitate to rely upon it as a defense.[16]

Current copyright law inhibits technological innovation for many of the same reasons.[17] In his 2013 article titled Copyright-Exempt Nonprofits: A Simple Proposal to Spur Innovation, Edward Lee, professor at Chicago-Kent College of Law, discusses at length the idea of the “Innovator’s Dilemma.”[18] Essentially, the more innovative or different a technology can be in terms of how it utilizes content, the more likely it is to face a copyright lawsuit or challenge.[19] This means that innovation in technology comes with an assumption of great risk of copyright liability.[20]

This assumption of risk also has a disparate impact on start-up companies.[21] The larger a company grows and the more users/customers it acquires the more attractive they are to copyright holders for a lawsuit.[22] However, a smaller company has far fewer resources to handle a drawn-out infringement lawsuit than a giant company.[23]  Therefore, start-up companies are caught in a catch 22 – either stay small and stifle their own innovation to avoid copyright suit, or further their own growth with the looming threat of a potentially bankrupting copyright suit if they get too big.[24]

The DMCA

The rapid advancement of technology in the last century has led to new challenges for legislators when balancing the interests of copyright holders with technological advancements.[25] In 1998, the Digital Millennium Copyright Act (DMCA) was passed to try and balance these interests for the digital world.[26] The DMCA is designed to protect copyright holders from online theft, specifically, to strengthen the control copyright holders have over the digital reproduction and distribution of their works.[27] Simultaneously, internet service providers are protected from liability when their users are engaged in infringing activity.[28] Title two of the DMCA creates “safe harbors” for online service providers against liability for user’s infringement given the service provider meets certain requirements.[29] The biggest of these requirements is that providers remove and block access to alleged infringing material when the copyright holder notifies them of an infringement claim.[30] 

The DMCA is seen as the “magna carta” of the internet, sketching – for better or for worse – the landscape of copyright law as it pertains to the internet.[31] However, the DMCA has faced immense criticism for failing to keep up with the landscape of rapidly developing technology on the internet.[32] For example, the “notice and takedown” model of the safe harbor requirements create a “guilty until proven innocent” system whereby simply alleging infringement is enough to take down a user’s content.[33] Due to this ease of getting content taken down, DMCA claims are the most often abused tactic copyright holders use when attempting to silence criticism and censor conversations regarding their works.[34]

III. Discussion

Artificial Intelligence

Artificial intelligence (AI) is shaping up to be the next great technological leap forward – akin to the advent of the internet, and it is facing a similar number of legal questions.[35] In legal terms, AI and AI companies have come under great scrutiny regarding the use of copyrighted materials.[36] This scrutiny is centered on three key issues: first, who owns the copyright over AI generated material; second, how copyrighted materials are used in the training of large language model AIs; and third, how to handle when AI generates copyright protected content.[37] While the first two issues are incredibly important when dealing with the landscape of AI development, and there are several articles that deal solely on these topics, therefore, this article’s scope will focus mainly on how to deal with AI generated content that is copyright infringing.

AI Generated Copyright Infringement

When AI is given a prompt to generate a piece of content, it responds by drawing on its dataset bank of billions of different words, texts, and images.[38] Significant amounts of this training consists of unlicensed copyrighted materials, and AI can potentially generate content that does not meaningfully differentiate from the original copyrighted work.[39] For example, Getty Images entered a high-profile lawsuit in 2023 when it learned Stability AI’s Stable Diffusion AI was generating images that Getty owned but with the watermark removed.[40] Getty Images is among other copyright holder plaintiffs which have faced difficulty using the DMCA against AI companies.[41] Getty Images as well as companies like GitHub and Raw Story have tried to say that the fact the AI models are removing copyright information from works creates a DMCA Article III issue.[42] However, the U.S. District Court for the Southern District of New York ruled that the removal of copyright information does not give companies standing to sue under Article III of the DMCA because there is no injury to plaintiffs in such a situation.[43] As of early 2025, many of these cases have yet to be resolved and neither Congress nor the Copyright Office have brought a legislative solution to these issues.[44]

Safe Harbor for AI Companies

The DMCA safe harbor provisions seem inadequate to help AI companies escape litigation.[45] The “notice and takedown” model works well enough for a platform like YouTube where users incur no cost for uploading and it is relatively simple for YouTube to remove content from their platform.[46] In contrast, the analog to AI would be to remove a copyrighted work from the dataset of AI and retrain it with a dataset that does not incorporate the copyrighted work.[47] Training an AI model can cost a company millions of dollars, and to do so every time a copyright holder wanted their work removed from an AI’s dataset would make the cost of operating an AI far too prohibitive.[48]

Balancing the Scales

So, if the DMCA is inadequately equipped to handle balancing copyright holder’s interests and the interests of AI companies, where do we go from here? The UK’s proposed solution is blanket amnesty for AI companies, but that tips the scales too far in one direction.[49]  T copyright system leaves much to be desired from both sides. Given the speed at which technology develops, case law may also not be the right solution for navigating this space. In 2023, the Copyright Office began a broad initiative to explore how copyright and AI can co-exist, but their conclusions cover much narrower issues regarding AI.[50] The Copyright Office’s conclusions came in two parts: part one was released in July 2024 and recommended legislative action to combat AI “digital replicas” more commonly known as “deep fakes.”[51] Part two concerns the copyrightability of AI works, and as such is outside of the scope of this article.[52]

The Copyright Office appears to have thrown its hands up in this space and requested legislative action to help combat AI’s copyright pitfalls.[53] Regardless of what the legislature does, it does not seem like an apt solution to continue with the framework provided by the DMCA. Just as the DMCA was introduced to adapt the Copyright Act to the emerging internet, a wholly new act may be needed to adapt both statutes to the world of AI.

IV. Conclusion

The constantly developing world of technology requires an equally developing legal framework to handle potential copyright violations. While the DMCA was able to adapt copyright law to the advent of the internet, it is proving inadequate as a method for balancing the interests of copyright holders and AI developers. A new, built from the ground up for AI, statute is required to adapt copyright law to balance the scales of various interests in copyrighted works and the future of American technology in AI.

---

[1] Sam Tabahriti, Musicians release silent album to protest UK’s AI copyright changes, Reuters (Feb. 25, 2025) https://www.reuters.com/lifestyle/musicians-release-silent-album-protest-uks-ai-copyright-changes-2025-02-25/#:~:text=LONDON%2C%20Feb%2025%20(Reuters),intelligence%20models%20using%20their%20work.

[2] Id.

[3] Id.

[4] Id.

[5] Digital Millennium Copyright Act (DMCA) § 103, 17 U.S.C. § 1201 (2006).

[6] Id.

[7] 17 U.S.C. § 101.

[8] 17 U.S.C. § 106.

[9] Id.

[10] U.S. Const. art. I, § 8, cl. 8.

[11] Id.

[12] Patricia Aufderheide et al., Copyright, Permissions, and Fair Use among Visual Artists

and the Academic and Museum Visual Art Communities, COLLEGE ART ASSOCIATION 5 (Feb.

2014).

[13] Id.

[14] Id.

[15] Id.

[16] Id.

[17] Edward Lee, Copyright-Exempt Nonprofits: A Simple Proposal to Spur Innovation (February 14, 2013). Chicago-Kent College of Law Research Paper No. 2013-08.       

[18] Id.       

[19] Id.

[20] Id.

[21] Id.

[22] Id.

[23] Id.

[24] Id.

[25] William Morriss, Will AI Destroy the DMCA Copyright Compromise?, Frost Brown Todd (Feb. 14, 2024) https://frostbrowntodd.com/will-ai-destroy-the-dmca-copyright-compromise/.

[26] Digital Millennium Copyright Act (DMCA) § 103, 17 U.S.C. § 1201 (2006).

[27] Id.

[28] Id.

[29] Id.

[30] Id.

[31] Jeffrey Cobia, The Digital Millennium Copyright Act Takedown Notice Procedure: Misuses, Abuses, and Shortcomings of the Process, Minnesota Journal of Law Science & Technology (2008).

[32] Id.

[33] Id.

[34] Id.

[35] Morriss, supra note 25.

[36] Id.

[37] Id.

[38] Id.

[39] Id.

[40]  McIntosh et. al., An End-of-Year Update to the Current State of AI Related Copyright Litigation, Ropes & Gray (Dec. 17, 2024) https://www.ropesgray.com/en/insights/alerts/2024/12/an-end-of-year-update-to-the-current-state-of-ai-related-copyright-litigation.

[41] Id.

[42] Id.

[43] Id.

[44] Id.

[45] Morriss supra, note 34.

[46] Id.

[47] Id.

[48] Id.

[49] Tabahriti, supra note 1.

[50] United States Copyright Office, Copyright and Artificial Intelligence Part 1: Digital Replicas (July 2024) https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-1-Digital-Replicas-Report.pdf.

[51] Id.

[52] United States Copyright Office, Copyright and Artificial Intelligence Part 2: Copyrightability (January 2025) https://www.copyright.gov/ai/Copyright-and-Artificial-Intelligence-Part-2-Copyrightability-Report.pdf.

[53] Id.