From AI to Posthumanism via ISO 42001 and AI Governance Standards

Abstract

The rapid diffusion of artificial intelligence (AI) into legal and compliance functions — from rule-based transaction monitoring to reinforcement-learning-aligned large language models (LLMs) capable of near-autonomous decision-making — necessitates governance frameworks adequate to the technology's operational reality. International standards and regulations, including ISO 42001:2023, the EU AI Act, and the guidance of the FCA, MAS, and FATF, aim to address this issue, yet a divergence persists between AI's operational reality and the frameworks intended to govern it — a compliance governance gap. This paper synthesises the literature across seven coded Themes — the history of AI in finance, RegTech and AML automation, AI/ML model development and explainability, regulatory supervision, generative AI and LLMs in legal compliance, AI risk detection, and robotic process automation — together with a separate, dedicated review of posthumanist theory's largely unexplored relevance to AI governance. On this basis, the paper develops a holistic analysis of how each Theme bears on AI-assisted compliance decision-making and concludes that AI has moved from tool to decision-make.  Using the Nvivo coding of around 50 SJR-Q1 papers on the topic published after 2022 when the AI-ChapGPT was born, it meaningfully increases effectiveness and efficiency by acting as a decision-maker within risk parameters that humans prescribe — but the ultimate decision matrix remains human. Read through a posthumanist lens, because humans already use AI as an inseparable extension of their own judgement, what ultimately governs outcomes is not the AI in isolation but the rules humans set for it: human and AI decision-making are not substitutable but interdependent, and effective governance requires them to operate hand in hand.