Securing agentic AI in banking: From governance gap to continuous control

Agentic AI is no longer experimental in banking - 65% of financial institutions are running AI in production, and 42% are deploying autonomous agents that execute transactions, route payments, and adjust credit parameters without human confirmation. But the security architecture most banks rely on was built for systems that responded, not systems that decided.

This webinar examines the structural security gap that emerges when AI systems operate across the behavior, API, and infrastructure layers simultaneously - and why perimeter-based defenses, quarterly policy reviews, and manual remediation workflows are fundamentally mismatched to the threat. We'll walk through the regulatory convergence happening across DORA, the EU AI Act, and NIST, the new accountability model hitting CISOs, CTOs, and CROs, and the architectural shift required to govern AI at the inference layer - where intent meets execution. Attendees will leave with a practical framework for closing the loop between adversarial testing, automated remediation, and real-time enforcement.

Key takeaways:

• The governance gap is structural, not staffing-related
• Agentic AI creates three simultaneous vulnerability layers
• The inference layer is the new security perimeter
• Regulatory convergence is real and accelerating
• Security, remediation, and enforcement must operate as a single continuous system