Agentic Artificial Intelligence and Systemic Financial Risk: A Complex-Systems Framework for Emerging-Market Supervision

Agentic artificial intelligence systems — models that plan, invoke tools, retain memory, and execute multi-step actions with limited human supervision — have moved from experimental pilots to production infrastructure inside major banks and trading firms during 2025-2026. This monograph argues that the systemic-risk consequences of this shift cannot be understood through financial regulation alone, nor through computer-science accounts of agent architecture alone: they require a complex-systems lens that treats interaction topology, cognitive opacity, and objective divergence among autonomous agents as structural properties of the financial network itself. [...] This is a conceptual and analytical work; it does not report new empirical estimation and should be read as a synthesis and framework-building exercise, not as a validated predictive model.