AI sits at the center of ICESCR Article 15’s most unresolved tensions. The article establishes three rights that have always existed in creative friction: the right to participate in cultural life, the right to enjoy the benefits of scientific progress, and the right to benefit from the protection of one’s own creative and scientific work. The AI economy has made that friction visible in a new way — and the United States navigates it without any international accountability framework.

What Article 15 Establishes

Article 15.1 names three distinct rights. First: everyone holds the right to take part in cultural life. Second: everyone holds the right to enjoy the benefits of scientific progress and its applications. Third: everyone holds the right to benefit from the protection of moral and material interests in any scientific, literary, or artistic production they authored.

Article 15.2 requires states to take steps for the conservation, development, and diffusion of science and culture. Article 15.3 requires states to respect the freedom indispensable for scientific research and creative activity — a provision that directly implicates IP regimes that could restrict research.

The Committee on Economic, Social and Cultural Rights issued General Comment 17 (2005) on Article 15.1.c, drawing a critical distinction: copyright law constitutes a legal mechanism that may protect the underlying human right, but the human right itself exists independently of copyright protection. A state could weaken or re-define copyright in ways that leave the underlying material interest right unprotected. GC 17 holds that the human right requires states to ensure authors receive meaningful material benefit from their work — not merely that some copyright system nominally exists.

Scientific Progress and Who Benefits

The “benefits of scientific progress” language in Article 15.1.b anticipates exactly the situation AI creates: a technology that represents genuine scientific advancement, whose practical applications produce enormous economic value, but whose distribution of benefit concentrates in a small number of entities rather than diffusing broadly.

CESCR General Comment 25 (2020) on science addresses this directly. Scientific progress carries obligations under the Covenant: states must ensure that applications of science benefit everyone and do not create or deepen inequalities. AI as a product of scientific progress falls within this framework — which asks not merely whether the technology exists, but whether its benefits reach the people whose work and data contributed to its development.

That question does not currently have an international accountability mechanism in the U.S. context. CESCR periodic review — the five-year cycle that ratifying states undergo — would require the United States to document how the benefits of AI development distribute across the population and what steps it takes to ensure equitable access to AI tools as applications of scientific progress.

Key question. Does scientific progress that concentrates its returns in a small number of companies satisfy the Article 15.1.b obligation that everyone enjoy its benefits? CESCR periodic review would require the United States to answer that publicly.

The Authorship Problem

Article 15.1.c creates a more direct collision with the current AI economy. Large language models and image generation systems trained on billions of creative works — text, art, music, code — in many cases without compensation to or consent from their authors. Whether that practice constitutes a copyright violation remains contested in ongoing litigation.

General Comment 17 establishes that the human right in Article 15.1.c does not collapse into copyright doctrine. Even if a court finds that training on copyrighted material falls within fair use, the question under the Covenant remains separate: do authors receive adequate material protection for the use of their work? A legal determination that training data use does not infringe copyright does not automatically satisfy the Article 15.1.c standard.

The United States has some of the world’s strongest IP law. But GC 17 asks a different question: does that law, as currently structured, ensure authors benefit materially from uses of their work that create significant economic value? In the AI training data context, for most authors, the current framework produces no compensation, no mandatory opt-out mechanism, and no revenue-sharing arrangement as a matter of right rather than voluntary industry practice.

CESCR review would require the United States to account for this gap. The absence of ratification means that review does not happen.

Cultural Participation and the AI Layer

Article 15.1.a’s right to cultural participation acquires new dimensions when cultural production routes increasingly through AI tools. Digital divides that prevent participation in cultural life — including the creation of cultural products — raise access questions the Covenant would require states to address.

General Comment 21 (2009) on cultural participation establishes that states hold an obligation to ensure access to the means of cultural expression, not merely to remove formal legal barriers. As AI tools reshape what cultural production looks like — who can create, at what cost, using what infrastructure — the adequacy of access measures requires examination.

The same broadband and device gaps that create unequal access to digital education (see Article 14) shape who can use AI-assisted creative tools. Cultural participation has historically correlated with access to instruments: printing presses, cameras, recording equipment, computers. AI tools represent the current iteration of that pattern. GC 21 asks whether the state takes adequate steps to ensure access, not merely whether the tools theoretically exist in a free market.

What Ratification Would Require

Ratification would not mandate specific outcomes on AI copyright, training data compensation, or AI tool access. CESCR does not function as a court.

What ratification would create: a periodic accountability mechanism. A five-year review cycle in which the United States would document how it balances the Article 15.1.c material interests of authors against the Article 15.1.b right to benefit from scientific progress; how it ensures AI-era scientific applications distribute benefits equitably; and what access measures it takes to ensure cultural participation in an AI-mediated cultural economy.

The U.S. debate over AI and authorship currently proceeds entirely through copyright litigation, executive orders, and voluntary industry commitments. Article 15 asks whether that framework adequately protects a human right — not a legal mechanism, but the underlying interest the drafters designed the mechanism to protect. The U.S. has not signed on to asking that question.

What You Can Do

The action guide covers how to contact your senators. Ratification requires a two-thirds Senate vote — a threshold that requires constituent pressure to move. Article 15 speaks to one of the most contested questions in the current technology landscape: who owns the benefits of AI, and what do creators owe and receive? The Covenant does not answer those questions definitively. It creates a framework in which the United States would have to answer them publicly, with evidence, on a regular cycle.

Part of the ICESCR Article Series — examining each of the treaty’s substantive articles through the lens of AI economic displacement.

EPISTEMIC FLAGS

  • General Comments 17 (2005), 21 (2009), and 25 (2020) cited from knowledge base; specific paragraph references have not undergone verification against official OHCHR text
  • The claim that AI training data use typically produces no compensation to authors reflects current industry practice as described in public reporting; ongoing litigation may alter this landscape
  • CESCR GC 25 (2020) on science and economic, social and cultural rights provides the cited “benefits distribution” framework; CESCR has not yet formally addressed the specific application to AI
  • The characterization of the Article 15.1.c standard as separate from copyright doctrine reflects the GC 17 analytical framework; international IP law scholarship continues to contest this distinction

Published by unratified.org · CC BY-SA 4.0