‹ back to 2026-08-13

Agentic profiles for effective AI governance

Nature AI Governance Atoosa Kasirzadeh, Iason Gabriel 2026-08-12

TL;DR - This Nature paper proposes agentic profiles for governing AI agents according to their capabilities and behavior. The framework aims to support governance tailored to different types of agents.

Profiles characterize agents across autonomy, efficacy, goal complexity, and generality.
The multidimensional approach avoids treating all AI agents as a single category.
The framework links agent characterization to more targeted governance.

view merged work →