Preserving attribution and accountability in AI-scale systems

Abstract Contemporary AI-mediated information systems operate at scales that can increasingly undermine the preservation of authorship continuity and temporal priority. While debates in AI governance, ethics, and accountability often presuppose the availability of stable attribution, this paper argues that such stability is a fragile infrastructural condition rather than a given. As information is copied, transformed, and recombined through automated systems, the linkage between claims, authors, and moments of assertion can progressively erode. This paper examines attribution collapse as a foundational socio-technical problem amplified by AI-scale systems and asks how authorship and temporal priority can be preserved without reliance on centralized authority, enforcement mechanisms, or prior adjudication. Adopting a conceptual and infrastructural approach, it identifies recurring failure modes in existing attribution practices and the design requirements for provenance mechanisms that can persist under conditions of automation, transformation, and scale. On this basis, the paper introduces BlockClaim as a minimal provenance framework designed to preserve attribution continuity and temporal integrity while remaining neutral with respect to truth, ethics, and enforcement. Rather than proposing a comprehensive governance solution, the framework is positioned as an enabling condition for accountability, supporting downstream ethical, legal, and institutional processes by stabilizing the historical trace of claims. The paper concludes by reflecting on the implications of non-adjudicative provenance for philosophy of technology and AI-mediated social systems.