Between creators and intermediaries, artificial intelligence exposes historical asymmetries in copyright remuneration
- The debate on AI and copyright: who is in focus?
The idea that protecting creations stimulates innovation and guarantees remuneration for creators is one of the core promises of the copyright system. In practice, however, this relationship is far less direct than traditional narrative suggests.
The study Between Incentives and Costs: Rethinking Intellectual Property (Costa; Garrote; Ramos, 2025) showed that there is no consistent evidence that strengthening copyright protection generates more innovation or economic growth. The impact of protection depends on context and the balance between incentive and access: overly restrictive rules can, paradoxically, inhibit the diffusion of knowledge and concentrate benefits among a few economic actors.
The advancement of artificial intelligence reignites this debate within a new technological landscape, bringing with it an age-old question: who actually benefits from rules protecting creative work?
For decades, the discussion on copyright remuneration centered on publishers, record labels, and digital distribution platforms (Toledo, 2023; UK Parliament, 2024). In recent years, the issue of remuneration for the use of protected works in AI systems has made its way onto the agendas of public consultations, courts, reports, and legislative debates in Brazil and worldwide (United Kingdom, 2025; ChatGptiseatingtheworld, 2025; Lucchi, 2025; Barreto, 2025).
These discussions typically focus on two primary questions:
- Who should be recognized as the author when AI participates in the creation of a work?
- How should the use of protected works in training AI systems be handled?
It is on this second front that the debate on remuneration gains momentum. Generative AI models rely on text and data mining – a technique that analyzes large content databases to “learn” patterns and generate new creations (Ribeiro; Garrote, 2025, p. 5). When these databases include protected works, the question arises: how should rightsholders be compensated?
Bill No. 2,338/2023 proposes that companies using protected content to train AI systems must remunerate copyright holders. In practical terms: if an AI model is trained on music, text, images, or videos, whoever holds the economic rights to that content should be paid for that use.
Yet there is an under-explored angle in this discussion: by remunerating the use of protected works, are we truly valuing creators, or merely reinforcing established structures of intermediation?
- Authorship vs. ownership: who benefits from copyright
When discussing “remunerating authors and creators” for the use of their works in AI systems, people often imagine that the money goes directly to whoever composed the song, wrote the text, or created the image. In practice, that is not how it works.
Brazilian legislation distinguishes two fundamental dimensions of copyright:
- Moral rights: tied to the creator’s identity, the right to be recognized as the author and to preserve the integrity of the work.
- Economic rights: related to the commercial exploitation of the work – reproduction, sale, adaptation, or distribution. These can be transferred or licensed.
Remuneration models are built upon economic rights. Therefore, payments for using works in AI systems tend to go to whoever holds these economic rights, not necessarily to whoever created the work.
In short: copyright exists to protect copyright holders, who may or may not be the actual creators of the works, since remuneration rights may have been contractually assigned to third parties. This distinction is critical to the debate on generative AI.
- The copyright remuneration chain
Between the creator and the public lies a long chain of intermediaries. Each stage – publishing, production, distribution, collective management, exhibition – involves actors who receive a portion of the value generated by the work. The outcome is predictable: most of the revenue does not reach creators directly, but rather goes to those who mediate access and control the rights.
Sectoral examples
- Journalism: staff journalists may assign their economic rights to the media outlet (Portal dos Jornalistas, 2021; Da Silva, 2025).
- Publishing industry: writers receive, on average, between 8% and 12% of the cover price (Pauquette, 2024; LiteraturaBR, 2025).
- Music: in streaming, around 30% of revenue stays with the platform, 55% goes to the record label, and 13% goes to the artist (Verscht; Anwer, 2023) – the so-called value gap (Litman, 2018).
Even collective management systems, such as Ecad, reproduce this inequality. In 2024, Ecad collected BRL 1.8 billion, of which 85% was distributed to rightsholders and 15% retained for administrative costs and member societies. The average annual income per rightsholder stood at BRL 4,200 – a small fraction of the average income of workers in the cultural sector (IBGE, 2023).
- Generative AI as a mirror of asymmetries
Generative AI emerges as yet another link in a chain already marked by inequalities. Proposals to remunerate the use of protected works in AI training are presented as ways to value creators; however, within a system where economic rights are concentrated among intermediaries, history is bound to repeat itself.
As Mantegna (2024) highlights, the current system protects economic rightsholders far more than creators themselves. Expanding this logic to AI risks amplifying existing inequalities.
Recent cases illustrate this movement. Financial Times, Axel Springer, and Le Monde have already signed licensing agreements with OpenAI, the company behind ChatGPT, to license their journalistic archives (Reuters, 2024). These negotiations benefit corporations holding vast catalogs, not individual creators.
Even the compensation models studied by Ducru et al. (2024) reinforce this trend. Concepts such as “pay-to-train”, “windfall profits”, or “AI royalties” run into technical and structural limitations. In almost every scenario, those who benefit most are owners of large catalogs, not the artists who created the original works.
No current model solves the problem of inequality in copyright remuneration.
AI merely reflects, on a new scale, the historical distortions of the cultural market.
- Copyright in Brazil: an outdated Law
Law No. 9,610/1998 was conceived in a pre-digital context. It does not provide for specific exceptions for data mining or mechanisms for proportional remuneration in digital environments.
Reform attempts date back to the 2007 National Copyright Forum (Intervozes, 2010), but without concrete progress. Bill No. 2,370/2019 sought to modernize issues such as orphan works, contractual assignments, and limitations for educational use, introducing the principle of “equitable remuneration.” Even so, the proposal stalled.
Other countries have made further progress. In the US, termination rights allow authors to reclaim their rights after 35 years of assignment (Sumner, 2024). Still, studies show that strengthening the legal framework alone is not enough. Kretschmer (2012) demonstrated that increasing copyright protection does not, on its own, improve creators’ income – since concentration occurs within market structures, not within the law itself.
6. Future Outlook
The regulatory dispute surrounding generative AI does not create new inequalities; it merely highlights those that already existed.
Remuneration chains in music, literature, and journalism reveal a consistent pattern: most of the revenue generated by creative works fails to reach the creators.
Brazil lacks data and transparency regarding these financial flows. Understanding how money moves among authors, performers, and intermediaries is essential for rethinking public policies and business models.
As Moody (2022) points out, the crisis is global: writers, musicians, and artists earn less even in highly regulated markets. In the context of AI, persisting with traditional copyright remuneration logic may simply result in digitalizing inequalities, rather than resolving them.
References
BRASIL. Lei nº 9.610, de 19 de fevereiro de 1998: altera, atualiza e consolida a legislação sobre direitos autorais e dá outras providências. Brasília, 1998. Available at: https://www.planalto.gov.br/ccivil_03/leis/l9610.htm
BRASIL. Projeto de Lei nº 2.338, de 2023: dispõe sobre o desenvolvimento, o fomento e o uso ético e responsável da inteligência artificial com base na centralidade da pessoa humana. Brasília, DF: Câmara dos Deputados, 2023. Available at: https://www.camara.leg.br/proposicoesWeb/fichadetramitacao?idProposicao=2487262
BRASIL. Projeto de Lei nº 2.370, de 2019: atualiza e consolida a legislação sobre direitos autorais. Brasília, DF: Câmara dos Deputados, 2019. Available at: https://www.camara.leg.br/proposicoesWeb/fichadetramitacao?idProposicao=2198534
BARRETO, E. PL da IA gera divergências sobre remuneração de direitos autorais. Política Brasileira, 10 set. 2025. Available at: https://politicabrasileira.com.br/politica/pl-da-ia-gera-divergencias-sobre-remuneracao-de-direitos-autorais/
CHATGPT IS EATING THE WORLD. Map of AI copyright lawsuits. 2025. Available at: https://chatgptiseatingtheworld.com/category/map-of-ai-copyright-lawsuits/
DUCRU, P; RAIMAN, J; LEMOS, R; GARNER, C; HE, G; BALCHA, H; SOUTO, G; BRANCO, S; BOTTINO, C. AI royalties: an IP framework to compensate artists & IP holders for AI-generated content. Cornell University, 5 abr. 2024. Available at: https://arxiv.org/pdf/2406.11857
ECAD – Escritório Central de Arrecadação e Distribuição. Relatório anual 2024. Rio de Janeiro: ECAD, 2025. Available at: https://www4.ecad.org.br/wp-content/uploads/2025/04/Relatorio-anual-versao-mercado-2025.pdf
INSTITUTO BRASILEIRO DE GEOGRAFIA E ESTATÍSTICA (IBGE). De 2011 a 2022, setor cultural ganha empresas, mas perde participação na economia. Agência de Notícias IBGE, Editoria: Estatísticas Econômicas, 1 dez. 2023. Available at: https://agenciadenoticias.ibge.gov.br/agencia-noticias/2012-agencia-de-noticias/noticias/38501-de-2011-a-2022-setor-cultural-ganha-empresas-mas-perde-participacao-na-economia
INTERVOZES – Coletivo Brasil de Comunicação Social. Ato pede urgência na reforma da Lei de Direitos Autorais. Intervozes, 31 maio 2010. Available at: https://intervozes.org.br/ato-pede-urgencia-na-reforma-da-lei-de-direitos-autorais/
KRETSCHMER, M. Does copyright law matter? An empirical analysis of creators’ earnings. Glasgow: University of Glasgow, School of Law, 2012. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2063735
LITERATURABR. Direitos autorais: quanto uma editora paga ao autor por seu livro? LiteraturaBr, 29 jan. 2025. Available at: https://www.literaturabr.com/2025/01/29/direitos-autorais-quanto-uma-editora-paga-ao-autor-por-seu-livro/
LITMAN, J. What we don’t see when we see copyright as property. The Cambridge Law Journal, Cambridge: Cambridge University Press, v. 77, n. 2, p. 536–559, 23 ago. 2018. Available at: https://www.cambridge.org/core/journals/cambridge-law-journal/article/what-we-dont-see-when-we-see-copyright-as-property/C15FED76D1ADF4A0CA1F526B95BE9A88
LUCCHI, N. Generative AI and copyright: training, creation, regulation. European Parliament, 2025. Available at: https://www.europarl.europa.eu/RegData/etudes/STUD/2025/774095/IUST_STU(2025)774095_EN.pdf
MANTEGNA, M. ARTificial: why copyright is not the right policy tool to deal with generative AI. The Yale Law Journal, v. 133, 22 abr. 2024. Available at: https://www.yalelawjournal.org/
MOODY, G. Creators everywhere are struggling, copyright is failing them: time to find something better. Walled Culture, maio 2022. Available at: https://walledculture.org/creators-everywhere-are-struggling-copyright-is-failing-them-time-to-find-something-better/
PAUQUETTE, Brad. How much money does a publisher make per book they sell? Brad Pauquette Blog, 4 dez. 2024. Available at: https://bradpauquette.com/2024/12/how-much-does-a-publisher-make-on-books-they-sell/
PORTAL DOS JORNALISTAS. Jornalistas querem retomar direitos autorais monopolizados pelas empresas. Portal dos Jornalistas, 9 abr. 2021. Available at: https://www.portaldosjornalistas.com.br/jornalistas-querem-retomar-direitos-autorais-monopolizados-pelas-empresas/
REINO UNIDO. Department for Science, Innovation and Technology. Copyright and AI: consultation. Presented to Parliament by the Secretary of State for Science, Innovation and Technology by Command of His Majesty. London: His Majesty’s Stationery Office, dez. 2024. Available at: https://www.gov.uk/government/consultations/copyright-and-artificial-intelligence/copyright-and-artificial-intelligence
RICARDO, J; GARROTE, M; RAMOS, P. Entre incentivos e custos: repensando a propriedade intelectual. Observatório de Direitos Autorais e Tecnologia. Microanálises Reglab, n. 2. Reglab, 2025. Available at: https://observatorio.cc/entre-incentivos-e-custos-repensando-a-propriedade-intelectual/.
DA SILVA, L. Direitos autorais de obras criadas por colaboradores e prestadores de serviço. Migalhas, 1 abr. 2025. Available at: https://www.migalhas.com.br/depeso/427404/direitos-autorais-de-obras-de-colaboradores-e-prestadores-de-servico
SUMNER, M. Top 7 recent changes to US copyright law. ScoreDetect Blog, 24 jun. 2024. Atualizado em: 6 maio 2025. Available at: https://www.scoredetect.com/blog/posts/top-7-recent-changes-to-us-copyright-law
TOLEDO, M. Entidades da música e do audiovisual se manifestam sobre falta de acordo com as plataformas sobre direitos autorais. Tela Viva, ago. 2023. Available at: https://telaviva.com.br/
UK PARLIAMENT. House of Commons. Culture, Media and Sport Committee. Creator remuneration: fifth report of session 2023–24. London: UK Parliament, 10 abr. 2024. Available at: https://publications.parliament.uk/pa/cm5804/cmselect/cmcumeds/156/report.html
VERSCHT, J; ANWER, A. Economics of streaming & the rise of the music artists’ rights and compensation. International In-house Counsel Journal, v. 16, n. 64, out. 2023. Available at: https://www.iicj.net/paper/1792
REUTERS. OpenAI to use FT content for training AI models in latest media tie-up. Reuters, 30 abr. 2024. Available at: https://www.reuters.com/technology/financial-times-openai-sign-content-licensing-partnership-2024-04-29/#:~:text=The Financial Times has signed,opens new tab startup
RIBEIRO, N.; GARROTE, M. Mapeamento Global de Leis de Direitos Autorais e Exceções para Mineração de Dados. Contexto Reglab. n. 1. Reglab, 2025.
