Abstract
Abstract Generative artificial intelligence (AI) has disrupted institutions built around relatively stable relationships among creator, work, publisher, audience, and law. Literature, music, and academia now confront production systems in which humans may conceive, prompt, edit, curate, verify, transform, or merely select machine-generated outputs. Framing this transformation as a simple conflict between “human” and “AI” creation is analytically inadequate. The deeper challenge is institutional: systems of authorship, copyright, peer review, quality assurance, disclosure, contractual representation, and professional accountability were not designed for environments in which provenance may be distributed across humans and machines. This article develops an Evaluation, Testing, Validation, and Deployment (ETVD) framework for AI-mediated creative and scholarly production. It argues that governance should not rely principally on AI detection, given evidence of false positives, false negatives, and uneven performance across linguistic populations. Instead, legitimate governance should integrate provenance, substantive quality, integrity, human responsibility, disclosure, and domain-specific risk. The article proposes a Provenance–Quality–Integrity–Accountability (PQIA) framework, complemented by a five-level AI involvement taxonomy and a four-stage ETVD cycle. Legitimacy, it argues, should depend not on proving that every element originated in a human mind, but on whether individuals or institutions presenting AI-mediated work can substantiate its provenance, defend its quality, disclose material AI involvement, verify its claims, and accept responsibility for its consequences. The central question is therefore not whether AI should enter creative and scholarly production, but under what conditions AI-mediated work can be trusted, attributed, evaluated, and legitimately accepted.



![Author ORCID: We display the ORCID iD icon alongside authors names on our website to acknowledge that the ORCiD has been authenticated when entered by the user. To view the users ORCiD record click the icon. [opens in a new tab]](https://www.cambridge.org/engage/assets/public/coe/logo/orcid.png)