Working Groups: Exploring the next frontier of agentic-AI in financial supervision
Working groupsRisk & compliance
2026-04-01 | 12:00 PM - 01:00 PM
Information
Regulators and supervisors are under pressure to move beyond chatbots and search toward agentic AI that can plan, act, and collaborate across supervisory workflows without compromising confidentiality. Recent supervisory surveys and guidance stress three imperatives: (1) governance‑first adoption (BIS/FSB), (2) privacy‑preserving architectures (federated learning, on‑prem/GPAI controls), and (3) measured autonomy with human oversight (BoE/FCA).
This session walks through live use cases implemented by authorities worldwide and debates privacy, confidentiality, and geopolitics of large models under evolving regimes. We’ll also show how agentic workflows can enable “straight‑through supervision” from data collection and validation to real‑time risk signals. Attendees will leave with practical patterns to deploy AI agents safely, and a checklist for readiness (data governance, model risk, third‑party dependency, and post‑market monitoring).
