New paper maps governance challenges for agentic AI's 2025 boom
As 2025 marked the 'Year of Agentic AI', researchers call for targeted oversight.
Deep Dive
A systematic review by Mubarak Raji and Masooda Bashir examines emerging governance challenges for agentic AI—systems that autonomously plan and execute tasks. The paper identifies features that distinguish agentic AI from traditional systems, synthesizes prevailing governance priorities, and lays the preliminary groundwork for developing a structured roadmap for responsible oversight. Presented at AIR-RES 2026, it provides an early scholarly foundation for adaptive governance frameworks.
Key Points
- 2025 saw rapid deployment of agentic AI systems that autonomously plan and execute tasks, moving beyond generative AI.
- The paper systematically reviews governance literature and identifies unique features (autonomy, goal-setting, tool use) that require targeted regulation.
- Authors propose a preliminary roadmap with priorities like transparency, accountability, and stakeholder roles for responsible oversight.
Why It Matters
As agentic AI takes autonomous actions, this review provides essential groundwork for governance frameworks that can prevent real-world harm.