International regulators are raising alarms over the rapid rise of "agentic" AI within the financial sector, warning that these highly autonomous systems could destabilize markets. In a recent report, the Financial Stability Board (FSB) emphasized that because these AI agents can plan and execute tasks with minimal human intervention, they introduce unique vulnerabilities. These include potential cybersecurity breaches, unintended illegal actions, and the risk that AI could deviate from corporate objectives faster than human supervisors can respond. With over half of surveyed financial firms already experimenting with or actively deploying these autonomous tools, the FSB is urging leadership teams to prioritize rigorous safeguards.

To manage these emerging threats, the FSB has proposed a framework of "sound practices" that encourages firms to treat AI agents more like synthetic employees while enforcing strict operational boundaries. The guidelines suggest mandating human approval for high-stakes decisions, such as large financial transactions, to ensure that technology remains under control. These recommendations, which are currently open for public comment until July 22, serve as a critical reminder that while AI offers significant efficiency in fraud detection and customer service, it must be governed by clear, enforceable protocols to prevent rapid-fire disruptions in the global financial system.