Artificial intelligence (AI) is undergoing a significant transformation in finance, evolving from a tool used primarily for market analysis to one that actively participates in trading decisions. This shift in functionality was a key topic at the recent Agentic AI and Automation in Finance Summit in Atlanta, where industry leaders convened to discuss the implications of AI’s expanding role in financial markets.
Historically, AI’s contributions to finance have revolved around predictive analytics, helping investors forecast price movements, identify trading patterns, and optimize strategies. However, as AI systems become more integrated into execution and decision-making workflows, the landscape of financial risk is becoming increasingly complex. As asset managers and financial infrastructure providers leverage AI more extensively, a critical question arises: what happens when an AI system generates incorrect outputs?
This concern was underscored during a panel discussion featuring Kaushal Sheth of GFT Technologies and Juan Mendez of BlackRock. The dialogue emphasized that the new category of operational risk introduced by AI extends beyond simple model performance metrics. The focus has shifted towards the trustworthiness of AI-generated outputs when applied to real-world market conditions.
AI systems, particularly those designed for autonomous operation across various stages of financial workflows, are increasingly capable of analyzing vast datasets, generating insights, and even executing trades without direct human supervision. This enhancement in efficiency brings with it a compressed margin for error. In traditional trading models, a flawed signal could be ignored or filtered out. Conversely, in autonomous systems, an error can propagate through to execution, posing substantial risks.
Kaushal Sheth highlighted the importance of understanding AI behavior under abnormal market conditions, which are often marked by regime shifts—periods when historical correlations break down and typical liquidity patterns evaporate. These challenging environments are where AI systems will be rigorously tested, yet they are also among the most difficult to simulate.
This gap between model development and real-world deployment presents a unique challenge for financial institutions. Although models can be trained on extensive datasets, their ability to perform in unpredictable market conditions remains uncertain. Financial firms are thus compelled to reassess how AI is woven into their core systems. The emphasis is shifting toward controllability, transparency, and resilience rather than merely performance metrics. Institutions are realizing that systems must not only yield robust results but also fail in predictable and manageable ways.
Sheth’s work at GFT Technologies and his involvement with Otonomii emphasize constructing AI architectures that achieve a balance between autonomy and necessary oversight. This ongoing evolution reflects a broader trend in the financial sector where the competitive edge will hinge on effective governance of AI models rather than merely the sophistication of those models.
As AI continues to penetrate deeper into finance, the inherent risks associated with incorrect outputs are becoming more pronounced. The challenge extends beyond simply being wrong; it encompasses the need to understand the subsequent ramifications of those errors in dynamic and often volatile market conditions. Financial institutions will need to cultivate systems that not only excel in stable environments but also demonstrate resilience in the face of uncertainty.
The integration of AI into financial markets is poised to radically redefine how risks are perceived and managed. As AI’s role evolves, the capacity to govern these systems effectively will become critical in navigating the complexities of modern finance. The future will demand a careful balancing act between leveraging AI’s capabilities while ensuring that its decisions can be trusted in real-time trading environments.
See also
AI Banking Keynote Highlights: Scott Steinberg on Personalization, Automation, and Ethics
Greenwich Finance Series Hosts 40+ Teens for AI and Financial Literacy Workshop
OpenAI Acquires TBPN in Major Media Move Amid $122 Billion Funding Round
Deloitte’s Reports Show Divergent AI Strategies in HR and Finance for 2026
AI Enhances Anti-Money Laundering in Crypto: Tackling $10B Financial Crime Challenge
