Group 1 - Financial institutions are rapidly transitioning AI agents from experimentation to production for various applications such as onboarding, fraud triage, transaction monitoring, and customer communication [1] - Model risk teams are facing increased pressure to validate more models frequently, necessitating built-in governance and evaluation for safe scaling of agentic AI [1][2] - The real concern in regulated financial environments is the potential for AI agents to "hallucinate," leading to unpredictable outcomes that differ from traditional deterministic systems [2][3] Group 2 - Core banking systems and compliance workflows rely on predictable logic, while agentic systems produce variable results, complicating testing and certification processes [3] - In financial compliance, inaccuracies in AI-generated reports can lead to significant control failures, which must be defensible under model risk management expectations [4] - There exists an "autonomy accountability gap," where the adoption of autonomous systems outpaces the development of accountability frameworks [5] Group 3 - Governance in organizations is often treated as an afterthought, with a focus on performance before addressing monitoring and oversight [6]