Position Paper: Model Access should be a Key Concern in AI Governance
Edward Kembery, Ben Bucknall, Morgan Simpson
TL;DR
The paper addresses the governance of how different styles of access to AI systems are provided and to whom, arguing that downstream harms and benefits hinge on access decisions. It introduces Model Access Governance as a formal research field, defining core elements (Model Aspects, Access Styles, Access Groups) and outlining a framework for policies, norms, and institutions. It then offers four concrete recommendations—extending AI evaluations to diverse access styles, industry commitment to responsible access, government capacity-building, and international consensus-building—along with six open problems to guide future work. The work highlights the need for empirical data and cross-stakeholder collaboration to enable risk-proportionate, evidence-based access decisions and to coordinate governance across borders.
Abstract
The downstream use cases, benefits, and risks of AI systems depend significantly on the access afforded to the system, and to whom. However, the downstream implications of different access styles are not well understood, making it difficult for decision-makers to govern model access responsibly. Consequently, we spotlight Model Access Governance, an emerging field focused on helping organisations and governments make responsible, evidence-based access decisions. We outline the motivation for developing this field by highlighting the risks of misgoverning model access, the limitations of existing research on the topic, and the opportunity for impact. We then make four sets of recommendations, aimed at helping AI evaluation organisations, frontier AI companies, governments and international bodies build consensus around empirically-driven access governance.