Risks of Artificial Intelligence Treasury Systems in Terms of Infrastructure and Liquidity
Deutsche Bundesbank has warned that AI treasury agents using similar models could exhibit herd behavior and create systemic risks in financial markets.
The shift of companies toward AI-powered agents for cash management and liquidity decisions brings new and major risks to the global financial system due to shared data sources and models.
AI-Powered Treasury Operations
Treasury departments are transitioning to artificial intelligence agents that can independently handle cash management, liquidity, and payment preferences.
While these systems provide economic benefits for individual financial managers, they pose serious risks across the broader financial system.
Shared Models and Herd Behavior Risk
In its September Monthly Report, the Deutsche Bundesbank warned that AI agents using similar models and data sources could make the same decisions simultaneously.
The transformation of individual optimization efforts into collective herd behavior can increase liquidity requirements and strain technical infrastructure.
Fast Reaction Times and Systemic Impacts
While traditional automation follows predetermined rules, agent systems interpret conditions to maximize returns or preserve liquidity.
The operation of software without delays in human approvals narrows reaction times and magnifies proactive systemic risks.
Concentration Risks and Proposed Solutions
The Bundesbank also drew attention to concentration risks around a limited number of foundational models and cloud providers, warning against infrastructure lock-ins.
To mitigate these risks, experts recommend clear authorizations, reliable authentication, monitoring mechanisms, and human intervention for anomalous situations.