In January, the Trump administration issued an executive order aimed at developing an artificial intelligence plan “to maintain and improve the global dominance of AI America.
Last month, the US government invited the public to share its ideas for an artificial intelligence action plan through Web Site of the Federal Register until March 15.
Last week, the American Association of Record Industry (RIAA) and nine other industry bodies published the content of their joint submission to the Office of Scientific and Technological Policy, featuring a number of proposals under the artificial intelligence action plan.
The application was submitted on behalf of music organizations, including NMPA; NSAI; AI2M; the American Federation of Musicians of America and Canada; the Alliance on the Rights of Artists; the Department of Professional Employees, AFL-CIO; the International Alliance of Theater Stage Employees; the Recording Academy; and the Screen Actors Guild-American Federation of Television and Radio Artists.
The document, submitted on March 14, 2025, goes far beyond a simple list of requirements. Organizations argue their stance on a fundamental principle they believe is essential for the future of the creative industry: “Progress in AI innovation and strong copyright protection are not mutually exclusive. This is not a zero-sum game.”
You can read the feed here.
Comments from music organizations sharply contrast with some views on AI that were recently shared by certain giants of the technology world.
As reported earlier this week, the creator of ChatGPT, OpenAI, in its own representation, called for fundamental changes to US copyright law that would permit artificial intelligence companies to use copyrighted material without permission or compensation to human rights holders.
Both OpenAI and Google have presented detailed political frameworks that could significantly affect music rights holders and other content creators.
Musical stars like Sir Paul McCartney, Paul Simon, and Bette Midler joined hundreds of Hollywood celebrities in signing a letter opposing these proposals.
“The gap left by AI companies using unlicensed copying to create competing models is wrong, just as it is wrong to undermine creators by utilizing their work without permission.”
Mitch Glazer, RIAA
Regarding the joint submission of RIAA, NMPA, and other industry bodies, RIAA chairman and executive director Mitch Glazer remarked: “One of the real practical barriers to adopting AI is the level of skepticism among end users towards AI…”
“To overcome such skepticism and build trust in AI, including assurance that training materials are legal and of high quality, the plan must require adequate accounting and transparency concerning AI training materials and the outcomes generated from AI algorithms.”
Glazer added: “America should prioritize the development of both AI and intellectual property. The exploitation of creators through unlicensed copying to produce competing works is indefensible. We hope the administration will support American creators and AI companies, enabling them to thrive in the global AI race.”
“American law protects creators, and we must insist that other nations do the same. American cultural products are the most sought after globally. We must firmly oppose any exceptions in their laws that would allow artificial intelligence firms to copy American content without compensation, as this could diminish the development of AI within the United States and lead to a double loss for our country.”
Key Proposals from the RIAA Submission
Here are the key proposals from the presentation by RIAA and others:
1. Licensing a free market for AI training
Sectoral bodies are advocating that developers of AI obtain proper licenses before using copyrighted materials to train their models. They cite examples such as OpenAI’s licensing agreements with media companies like Shutterstock and Financial Times as encouraging precedents.
The recommendation underscores that proper licensing for AI training fosters “symbiotic relationships” between rights holders and AI developers.
“(B)efore deploying AI systems, developers must obtain the appropriate licenses for any copyrighted material used to train their models, as well as permission if they utilize someone’s name, image, likeness, or voice in connection with such training.”
“In accordance with US laws and free market principles, these licenses should be negotiated without regulation to ensure fair value and economic competitiveness.”
2 Countering international exceptions for text and data mining
RIAA, NMPA, and other groups oppose international copyright exemptions that could facilitate the unauthorized use of American intellectual property.
They argue that such exemptions, often made prior to the rise of generative AI, threaten the copyright sector and undermine fair compensation for creators.
“(W)e monitor specific TDM exemptions, which could be justified only for very specific types of unprotected data, as some might exploit them as a means to access American creations for free, depriving US copyright owners of compensation and undermining the reliable licensed data market necessary for AI training,” they wrote in their submission.
“To ensure a level playing field, restrain investments in AI, and prevent foreign appropriation of American works and data, the US must manage this issue and resist any TDM exceptions that would impact copyrights.”
3 Strengthening current copyright laws
The statement asserts that existing US copyright law is sufficiently robust to address challenges posed by AI concerning copyright issues.
“In particular,” the statement notes, “the fair use doctrine prevalent in the US legal framework (which is unique to US law and should not and cannot be imposed on other nations lacking similar values and decades of judicial precedent) provides a controlled and nuanced approach to addressing liability for the unauthorized use of copyrighted materials for AI training.”
The submission also points to a recent Reuters ruling as a reaffirmation that the unauthorized use of copyrighted material for AI training constitutes a direct infringement, rather than fair use.
“The court confirmed that the impact of usage on existing and potential markets is the most crucial element of fair use, and there is clearly a potential market for the materials involved in AI teaching,” the submission continues.
The application emphasizes: “(O)ver decades, the legal precedents within this country have established that copyright is intended to protect only human expression. The recent copyright authoring law report confirms that this established doctrine remains applicable in the context of AI.”
“This approach aligns with our constitutional values and supports human prosperity. The Constitution protects the rights of individuals; machines must not and should not possess the same rights.”
4. Protecting voice and likeness rights
They emphasize the growing concerns regarding Deepfake technology, including potential misuse for fraud and personal operations.
Accordingly, they strongly support proposed legislation aimed at protecting individuals’ rights to voice and likeness from unauthorized AI replication.
“While unauthorized AI cloning of voices and likenesses has notably impacted the creative community, the use of AI technologies for such purposes carries far broader implications for personal and national security,” the organization notes.
The submission continues: “There are numerous unauthorized AI voice models and likenesses of various celebrities readily available, along with accessible AI voice models or images of political or corporate figures, such as President Trump, Vice President Pence, and Elon Musk.”
“Moreover, various services and applications have recently emerged online, allowing users to create unauthorized clones of voices or likenesses without requiring substantial technical knowledge.”
They also note that “in order to combat these risks, adhering to these principles, the Trump administration and the plan should support the proposed legislation relevant to impersonation.”
Such legislation would establish a national foundation protecting human rights regarding voice and likeness in relation to digital replicas, while providing victims substantial compensation for unauthorized digital uses, as long as appropriate First Amendment protections are maintained.
5 Promotion of AI transparency
In their submission, RIAA et al. recommend that artificial intelligence companies keep detailed records of training materials and provide clear summaries of works utilized for developing their AI models.
They endorse a proposed “train law” that establishes a judicial process for copyright owners to investigate potential unauthorized uses of their works.
As indicated in the submission, “To overcome skepticism and build trust in AI—including confidence that training materials are legal and high quality—the plan must mandate adequate records and transparency regarding training materials and the results produced by AI algorithms.”
“AI companies must furnish the public with reasonable reports detailing works used to train their AI systems.”
The submission adds: “In line with the National Institute of Standards and Technology (NIST), the risk management framework for artificial intelligence, companies should maintain sufficient records regarding the origins of training materials utilized to educate their systems.”
“As highlighted by NIST, “identifying the source of the training data and confirming attribution of AI systems to distinct segments of training data can enhance both transparency and accountability.”
“AI companies must provide the public with adequate reports specifying works used to train their AI systems. These public summaries should include enough information for copyright owners to ascertain whether their works were utilized without appropriate licensing for such use.”
Lucas focuses on the interaction between music and society. He specializes in how music influences and reflects cultures, analyzing the role of sound in social, political, and cultural events around the world.
