Dr Zeynep Engin, Co-Founder of The Digital Statecraft Academy, has published a new paper in the Journal of Responsible Technology introducing a novel framework for governing AI systems as they grow more autonomous and capable.
The paper, Human-AI Governance (HAIG): A Trust-Utility Approach, argues that current approaches to AI governance — which tend to sort systems into fixed categories such as “high risk” or “human-in-the-loop” — are poorly equipped to handle the reality of how AI actually evolves in deployment. As foundation models and multi-agent systems blur the lines between tool and partner, governance frameworks need to track continuous, shifting relationships rather than assign static labels.
The HAIG framework addresses this by mapping human-AI relationships across three dimensions — Decision Authority, Process Autonomy, and Accountability Configuration — each understood as a spectrum rather than a binary. It introduces a set of “trust thresholds”: critical points where governance requirements shift qualitatively, enabling organisations and regulators to anticipate challenges before they arise rather than respond after the fact.
Central to the framework is a trust-utility orientation that reframes governance not merely as a constraint on AI deployment, but as the condition under which human-AI collaboration can realise its full potential. This distinguishes HAIG from predominantly risk-based approaches: rather than asking only what could go wrong, it asks what governance arrangements are needed for AI systems to genuinely earn and sustain warranted trust across different contexts and stakeholders.
The framework also speaks directly to a challenge that regulators and deploying organisations frequently encounter but rarely name explicitly: governance asymmetry. AI systems often advance rapidly along one dimension — say, operational autonomy — while accountability structures lag behind. HAIG makes these misalignments visible and trackable, providing a diagnostic tool for organisations to assess where their governance arrangements are becoming misaligned with their technical deployments before those gaps become acute.
This research connects directly to the principles at the heart of The Digital Statecraft Manifesto. The Manifesto calls for dynamic and adaptive governance that moves beyond static regulation to anticipate technological change — precisely what HAIG is designed to enable. Its dimensional architecture embodies the principle of stewarding the human-machine partnership, providing structured tools for leaders to govern the hybrid environments where human judgement and machine intelligence now intersect. And by insisting that accountability must keep pace with autonomy, the framework gives practical expression to the Manifesto’s warning that as more decisions become automated, legitimacy becomes harder to obtain and maintain.
Ultimately, HAIG is a contribution to the third future the Manifesto envisions — Resilient Public Futures — in which networked, adaptive institutions harness AI in service of democratic values rather than allowing algorithmic advancement to outpace the governance structures meant to guide it. It offers public leaders, regulators, and deploying organisations a concrete analytical foundation for governing wisely at a moment when the stakes of getting this right could hardly be higher.
The paper is available open access in the Journal of Responsible Technology (Vol. 26, 2026).