Governing AI Execution as Institutional Infrastructure
This report summarizes discussions from a Risk.net AI governance roundtable held in New York in June 2026, in collaboration with Cisco. Key takeaways emphasize that AI governance is shifting from model approval to operational control as agentic tools enter workflow execution. Panellists highlighted the need for dynamic, live AI inventories rather than periodic registers, and stressed that accountability must follow the chain of action across business, risk, compliance, and technology silos. Controls must operate at run time, with access rights, telemetry, policy enforcement, and audit records embedded in workflows. Leaders foresee an operating model valuing human skills to interrogate outputs and control automation. The article notes that regulatory guidelines from the US Federal Reserve, OCC, DORA, and EU AI Act are raising the bar from evidence of controls to evidence of decision quality.
Editorial responsibility
- No named human review is recorded for this page.
- Reports are grouped by semantic similarity and deterministic rules. Language models may assist titles, summaries, translation and cross-source analysis; the page itself is projected from evidence records.
- Current automated evidence projection