Goldman Sachs and Deutsche Bank are testing agentic artificial intelligence systems to monitor trading activity in real time. The shift signals a move beyond rule-based alerts toward adaptive oversight designed to detect complex misconduct patterns earlier.
According to Bloomberg, both banks are exploring or deploying AI agents that analyze orders and executions as they occur, rather than relying solely on predefined thresholds. Deutsche Bank is working with Google Cloud to develop tools that review large datasets across venues and flag anomalies in near real time. Goldman Sachs is extending prior AI investments in trading and risk into compliance monitoring functions.
Can Agentic AI Reduce False Positives In Surveillance?
Traditional surveillance systems trigger alerts when trades breach size limits or deviate from benchmarks. But modern markets generate vast cross-asset data flows, often producing high volumes of false positives while missing subtler behavioral signals. The newer agentic systems are designed to examine multiple indicators simultaneously, compare activity against historical behavior, and escalate unusual combinations rather than isolated events. Could that materially reduce compliance noise without weakening scrutiny?
Bloomberg reports that the AI agents are not replacing human supervisors. Instead, they function as an additional analytical layer, surfacing cases that warrant manual review. In Deutsche Bank’s implementation, the tools can identify what the report describes as “complex anomalies” across order flow, timing, market conditions, and trader history. Compliance staff remain responsible for determining whether flagged conduct requires escalation or disciplinary action.
Regulators in the US and Europe require firms to maintain effective systems and controls against market abuse. While no mandate specifies agentic AI, banks face mounting pressure to handle growing data volumes with tighter oversight standards. If these systems demonstrate explainability and audit resilience, broader adoption across global trading desks may follow as institutions prepare for the next regulatory review cycle.
