Agentic AI Raises New Recordkeeping Questions for Government

Unlike traditional AI tools that primarily analyse information or generate content, agentic AI can plan, reason and take actions with greater autonomy. While this creates opportunities to improve services and reduce administrative workloads, it also introduces significant information governance questions around accountability, oversight and control.

For records and information management professionals, the development is particularly significant. When an AI agent can create, classify, move, modify or act on information across multiple systems, organisations need to determine what records must be captured, how actions are documented, what metadata and audit trails are required, and who remains accountable for decisions and outcomes.

Australia has already begun addressing these challenges through the Australian Government’s Agentic AI addendum, released in May 2026 as part of its broader framework for responsible government use of AI. The OECD working group will now enable Australia to share this experience with countries including Canada, Estonia, Japan, Singapore, South Korea, Israel and the United Kingdom.

For the R&IM profession, agentic AI reinforces an important message: information governance cannot be added after AI systems are deployed. Recordkeeping, accountability, provenance, access controls and lifecycle management need to be considered as part of the design and governance of these systems from the beginning.

Read more: Digital Transformation Agency, Australia joins new OECD working group on agentic AI in government, 29 September 2026.