Executive Summary:
Financial institutions are increasingly experimenting with agentic AI systems that can plan, initiate, and execute complex multi-step tasks with limited human intervention. These technologies have the potential to transform activities ranging from investment research and portfolio management to trading, compliance, and client service.
This paper examines the unique risks associated with increasingly autonomous AI systems and argues that governance frameworks must evolve before adoption reaches scale. The author identifies a broad range of risks, including operational, governance, cybersecurity, model, concentration, compliance, and systemic risks. A key concern is that highly autonomous systems may behave in ways that are difficult to anticipate, monitor, or explain using existing risk-management approaches.
The paper compares regulatory developments across Canada, the United States, and Europe and highlights gaps between the pace of technological innovation and current supervisory frameworks. It also proposes practical safeguards, including autonomy-based oversight, agent stress testing, security-by-design principles, and fiduciary controls.
For CEOs, boards, Chief Risk Officers, and senior executives, the paper offers a practical framework for assessing institutional readiness for increasingly autonomous AI systems. For regulators and policymakers, it highlights the importance of addressing governance and accountability challenges before widespread adoption creates new vulnerabilities.
The central message is straightforward: the question is no longer whether agentic AI will be used in finance, but whether institutions can develop the governance, controls, and oversight necessary to deploy it safely and responsibly.
Three Main Takeaways
- Agentic AI systems introduce new risks because they can make and execute decisions with increasing levels of autonomy, creating challenges that extend beyond traditional model and operational risk.
- Governance, oversight, and control frameworks should be established before agentic AI becomes deeply embedded in critical financial processes.
- Policymakers, regulators, and financial institutions must prepare for potential impacts on market integrity, financial stability, accountability, and cross-border supervision.
Special Issue on Emerging Risks by GRI in partnership with the Journal of Portfolio Management