The proliferation of generative artificial intelligence (AI) systems in academic environments has produced a profound normative vacuum regarding the attribution and ethical valuation of algorithmically co-produced scholarly texts. This paper introduces, formally defines, and theoretically develops the concept of Algorithmically Assisted Authorship (AAA) — understood as a hybrid modality of textual production in which academic output emerges from structured interaction between human cognitive agency and generative algorithmic systems, yielding distributed authorship and layered epistemic responsibility. The paper advances two central claims. First, that undeclared high-level AAA — particularly at AAA-Level 2 (Substantive Co-Generation) and AAA-Level 3 (Delegated Generation) — constitutes a form of epistemic misrepresentation: a distortion of authorship attribution systems rooted not in textual appropriation but in the concealment of algorithmic cognitive delegation within evaluative academic frameworks. This wrong is structurally distinct from traditional plagiarism and requires its own normative vocabulary and regulatory apparatus. Second, that the ethical implications of AAA are intensified in contexts characterized by algorithmic dependency and epistemic asymmetry, particularly those of the Global South. The paper proposes the Hybrid Algorithmic Transparency (HAT) Framework — comprising four principles: Declarative Transparency, Proportional Attribution, Retained Human Accountability, and Context-Sensitive Disclosure — as a normative instrument adaptable to diverse institutional, cultural, and geopolitical contexts.
Jhonatan Estiven Correa Londoño (Tue,) studied this question.