By: Jessica Payne, Volume 43 Staff Member

View/Download PDF Version: The TRAIN Act and Copyright (Payne)

Since generative artificial intelligence’s (AI) soar to popular use over the past several years, a slew of legal issues has risen demanding the attention of Congress. The past two Congressional sessions have featured a significant number of AI-related legislation aimed at controlling the content AI creates.[1] Of considerable note, the recent bipartisan supported NO FAKES (Nurture Originals, Foster Art, and Keep Entertainment Safe) Act provides right of publicity protection for all who may have their likenesses used without authorization.[2] This Act was bolstered by media and entertainment industry leaders for establishing “the first-ever federal right of publicity in the United States . . . provid[ing] a national standard to protect creators’ likenesses from being used without their consent.”[3] Congress has now turned its eye toward the substantive development of AI renderings themselves.[4] Generative AI analyzes data inputs in order to create its own outputs based on the information provided to the system.[5] However, legal issues arise within the AI administrator’s collection, input, and AI creation of derivative works based on copyrighted works.[6] Unlike trademarks, which are recognizable and distinctive branding, lettering, and other symbols, copyrighted works may be more difficult to recognize[7] – particularly as the AI rendition may be an amalgam of other copyrighted works.[8]

In November of 2024, Senators introduced the Transparency and Responsibility for Artificial Intelligence Networks (TRAIN) Act.[9] The TRAIN Act creates “an administrative subpoena process to assist copyright owners in determining which of their copyrighted works have been used in the training of artificial intelligence models.”[10] Under the Copyright Act of 1976, copyright requires that a creation be an original work of authorship fixed in a tangible medium.[11] Unlike a patent, a copyright does not need to be registered in order for the author to obtain a right to control their work.[12] As such, artists, writers, and digital content creators all create enforceable copyrights online[13] and physically that may be inserted into AI programs for learning and its own creation.[14] Imparted to copyright owners are a set of exclusive, alienable rights, including the right to create derivative works, reproduce the work, distribute copies, to display the work, and to perform the work.[15] The TRAIN Act, through the subpoena right, would allow the owner or transferee of a copyright to “obtain copies of the training material, or records sufficient to identify” that their original work was “used to train the generative artificial intelligence model.”[16] Much like the Digital Millenium Copyright Act (DMCA), the TRAIN Act requires that the copyright owner request a subpoena based on a “subjective good faith belief,” followed by a timely disclosure of any AI training materials by the developer.[17] The TRAIN Act would further benefit copyright holders, as the failure on the part of the developer to provide such information “shall provide a rebuttable presumption that the model developer or deployer made copies of the copyrighted work.”[18]

As generative AI continues to develop and increase in its use, the need for regulation becomes vital in order to protect creatives.[19] At this time, the Copyright Act provides no protection for non-human authorship.[20] This means that a major challenge for authors is that they must defend and assert the human, genuine, creativity they poured into a work. While the benefits of generative AI allow for faster processing and the analyzing and summarization of numerous sources of written text, this kind of legislation becomes even more critical to protect the likeness of public figures, visual arts, and music. Textual summary and analysis have historically been monitored and detected by similarly sophisticated plagiarism software.[21] However, seemingly harmless AI renditions of artistic works, music, and writing pose a much larger issue, particularly to lesser-known artists. For example, popular culture is already seeing several very explicit AI infringements or uses on copyright, like “Heart on My Sleeve,”[22] released using Drake and The Weeknd’s likenesses, and Getty Image’s current lawsuit against Stability AI.[23] However, these artists have already reached a caliber of recognition such that their voices, style, and cadence would likely be immediately recognizable when listening to an AI generated song – allowing them to ostensibly be aware when their likeness and copyrights are infringed upon.[24] Like many other copyright infringements, attributing a certain style or voice to a creator may damage one’s reputation or cultivated artistic imagery.[25] One of the primary benefits and aims of copyright law is the copyright holder’s ability to control nearly all uses of their work.[26] In allowing copyright holders to have full knowledge of uses of their material, they are able to fully able to take advantage of the rights vested in them through the Copyright Act.[27]

Ultimately, the NO FAKES Act and the TRAIN Act represent necessary legal advancement in the nexus between AI and copyright law in order to protect creators. The many advantages to generative AI are countered by the great impact that it may have on human creation and existing works.[28]  However, it is clear that there is much more to be considered by Congress, creatives, and AI system developers as this technology continues to become intertwined with copyright.

[1] See Artificial Intelligence Legislation Tracker, Brennan Center for Justice (Dec. 30, 2024) https://www.brennancenter.org/our-work/research-reports/artificial-intelligence-legislation-tracker.

[2] See NO FAKES Act, S. 4875, 118^th Cong. (2024).

[3] Nina Frazier, NO FAKES Act Introduced in the Senate: Protecting Artists’ Rights in the Age of AI, The Recording Academy (Aug. 9, 2024), https://www.recordingacademy.com/advocacy/news/no-fakes-act-introduced-in-the-senate.

[4] Ursula Perano, Congressional Leaders Negotiating Potential Lame-Duck Deal to Address AI Concerns, Politico (Oct. 18, 2024) https://www.politico.com/news/2024/10/18/congress-ai-schumer-00184430.

[5] See Generative AI Beginner’s Guide, Google Cloud (Jan. 13, 2025), https://cloud.google.com/vertex-ai/generative-ai/docs/learn/overview.

[6] Rachel Reed, ChatNYT, Harvard Law Today (May 22, 2024), https://hls.harvard.edu/today/does-chatgpt-violate-new-york-times-copyrights/ (discussing the New York Times’ lawsuit against OpenAI for providing portions of articles in its outputs).

[7] 15 U.S.C. § 1127.

[8] Id.; Reed, supra note 6 (“[T]he actual training was infringing, in the sense that when OpenAI scraped all this data from the web, it had to make copies . . . the actual large language model that results at the end of the training is either a copy or possibly a derivative work.”); Generative AI Beginner’s Guide, supra note 5; Trademark, Patent, or Copyright, U.S. Pat. & Trademark Off., https://www.uspto.gov/trademarks/basics/trademark-patent-copyright; Brian Moriarty, Timothy Meagher, & Daniel Fleisher, Digital Image Creation Using AI Risks Copyright Infringement, Bloomberg Law (Sept. 16, 2024), https://news.bloomberglaw.com/us-law-week/digital-image-creation-using-ai-risks-copyright-infringement, (“AI programs sometimes don’t create brand new images, but instead produce an image from the training set.”).

[9] Angela Yang, Senator Introduces Bill to Compel More Transparency from AI Developers, NBC News (Nov. 25, 2024), https://www.nbcnews.com/tech/senate-bill-transparency-ai-developers-rcna181724.

[10] TRAIN Act, S. 5379, 118th Cong. (2024).

[11] See 17 U.S.C. § 106.

[12] See Copyright Basics, U.S. Copyright Off. 5 (Sept. 2021), https://www.copyright.gov/circs/circ01.pdf.

[13] Id. at 4.

[14] Winston Cho, Some Good News for Hollywood Creators Suing AI Companies, Hollywood Reporter (Feb. 11, 2025), https://www.hollywoodreporter.com/business/business-news/reuters-ross-first-ai-copyright-case-lawsuit-1236133743/, (discussing a recent district court’s holding that the fair use exception in the Copyright Act does not apply to generative AI’s use of copyrighted material to create new outputs).

[15] See 17 U.S.C. § 106.

[16] TRAIN Act, S. 5379, 118th Cong. (2024).

[17] Id.; 17 U.S.C. § 512(a).

[18] TRAIN Act, S. 5379, 118th Cong. (2024).

[19] As generative AI continues to look to already created material to aid in its generative outputs, copyright holders will likely continue to face AI companies’ fair use arguments. Cho, supra note 14, (“[T]he court’s ruling will be ‘heralded by existing groups of artists and content creators as the key to their case against the other generative AI systems.”). Reed, supra note 6.

[20] Thaler v. Perlmutter, 687 F. Supp. 3d 140, 146 (D.D.C. 2023) (holding that copyright law is meant to protect human authorship as its central purpose and will not extend to non-human creations) (“Human authorship is a bedrock requirement of copyright.”); See Restatement of Copyright Law § 22, Reporter’s Note (d) (Am. L. Inst. Tentative Draft No. 2, 2022). However, it is also important to note that this has recently shifted based on guidance from the U.S. Copyright Office in a report released in January 2025. While the Copyright Office remains aligned with the need for human authorship, it also confirmed “that the use of AI to assist in the process of creation or the inclusion of AI-generated material in a larger human-generated work does not bar copyrightability. U.S. Copyright Off., PR 25-010, Copyright Office Releases Part 2 of Artificial Intelligence Report (Jan. 29, 2025), https://newsroom.loc.gov/news/copyright-office-releases-part-2-of-artificial-intelligence-report/s/f3959c36-d616-498d-b8f9-67641fd18bab.

[21] This is especially evident in recent efforts of higher education professionals seeking to disentangle plagiarism and AI usage in academic writing. Lauren Coffey, Professors Cautious of Tools to Detect AI-Generated Writing, Inside Higher Ed (Feb. 9, 2024), https://www.insidehighered.com/news/tech-innovation/artificial-intelligence/2024/02/09/professors-proceed-caution-using-ai.

[22] See Chloe Veltman, When You Realize Your Favorite New Song Was Written and Performed by…AI, Nat’l Pub. Radio (Apr. 21, 2023), https://www.npr.org/2023/04/21/1171032649/ai-music-heart-on-my-sleeve-drake-the-weeknd; See also Joe Coscarelli, An A.I. Hit of Fake ‘Drake’ and ‘The Weeknd’ Rattles the Music World, N.Y. Times (Apr. 24, 2023), https://www.nytimes.com/2023/04/19/arts/music/ai-drake-the-weeknd-fake.html; BUKO, Ghostwriter – Heart on my sleeve Drake ft. The Weeknd (Remastered AI Song) BUKO, YouTube (Apr. 20, 2023), https://www.youtube.com/watch?v=rQssjhX31Z0.

[23] See Getty Images (US) Inc. v. Stability AI, Inc., No. 1:23-cv-00135 3 (D. Del. Feb. 3, 2023).

[24] Samantha Murphy Kelly, Viral ‘BBL Drizzy’ AI Drake Diss Track Company Gets Sued, CNN (June 29, 2024), https://www.cnn.com/2024/06/24/tech/riaa-lawsuit-ai-copyright-infringement/index.html, (reporting that the Recording Industry Association of America (RIAA) filed a suit against the generative AI company behind “BBL Drizzy” for copyright infringement of the owned sound recordings of record labels that fall within their music trade group) (“In April, more than 200 artists . . . signed an open letter organized by the non-profit Artist Rights Alliance calling on AI developers . . . to ‘cease the use of artificial intelligence to infringe upon and devalue the rights of human artists.’”); Kristin Robinson, Metro Boomin’s ‘BBL Drizzy’ Is More Than a Joke – It Could Signal the Future of Sampling, Billboard (May 15, 2024), https://www.billboard.com/business/tech/metro-boomin-bbl-drizzy-future-ai-sampling-1235682587/; See Metro Boomin, BBL Drizzy BPM 150.mp3, SoundCloud (May 5, 2024), https://soundcloud.com/metroboomin/bbl-drizzy-bpm-150-mp3.

[25] Gil Appel, Juliana Neelbauer, & David A. Schweidel, Generative AI Has an Intellectual Property Problem, Harvard Business Review (Apr. 7, 2023), https://hbr.org/2023/04/generative-ai-has-an-intellectual-property-problem.

[26] This is primarily limited by the Fair Use doctrine which permits various uses of copyrighted material, primarily for news, education, and parody. 17 U.S.C. § 107.

[27] See 17 U.S.C. § 106.

[28] Appel, supra note 25. Geoff Brumfiel, Research Shows AI Can Boost Creativity for Some, But at a Cost, NPR (July 12, 2024), https://www.npr.org/2024/07/12/nx-s1-5033988/research-ai-chatbots-creativity-writing, (describing a new study in the United Kingdom about the impact of AI on creativity in creative writing, finding that the individuals benefitted creatively but that the group, collectively, received a lower creativity score).