Big banks are experimenting with a new kind of artificial intelligence designed to do more than just scan for keywords or follow fixed rules. Instead of relying solely on static alerts, some trading desks are starting to use systems that can reason through patterns in real time. These AI tools aim to spot suspicious trading behavior as it happens and flag cases for human review. Goldman Sachs and Deutsche Bank are leading the way in exploring or deploying these so-called “agentic” AI systems for trade surveillance.

Moving Beyond Traditional Surveillance

Traditional automated surveillance systems rely on preset rules, such as triggering an alert if a trade exceeds a certain size or deviates from a benchmark. Compliance teams then review these alerts manually. But with the massive volume of data generated across different markets and asset classes, these static rules often produce false positives or miss subtle forms of misconduct. Modern markets involve complex transactions, multiple time zones, and various trading venues, making it difficult for fixed rules to keep up.

Bloomberg reports that the new agentic AI systems are meant to go beyond these basic checks. Instead of just matching trades against a checklist, these AI agents analyze multiple signals and compare current activity with historical patterns. They look for unusual combinations of actions, timing, and relationships between trades, which can suggest manipulative behavior or other misconduct. These tools are not meant to replace compliance officers but to act as an extra layer of monitoring, helping to surface cases that need closer human inspection.

Deutsche Bank’s Collaboration with Google Cloud

Deutsche Bank is working with Google Cloud to develop these advanced AI agents. The goal is to review large volumes of order and execution data in near real time and identify anomalies quickly. This effort reflects how financial institutions are increasingly applying generative AI and large language models beyond simple chatbots. Instead of answering customer questions, these models analyze structured and unstructured data streams tied to trading activities.

The AI tools can help detect “complex anomalies” by examining relationships between trades, market conditions, timing, and trader history. This approach allows the system to identify patterns that might indicate misconduct but wouldn’t stand out through traditional rule-based checks. Human compliance staff remain responsible for reviewing flagged cases and deciding what action to take, but the AI helps make that process more efficient and accurate.

Goldman Sachs’ Investment in AI Surveillance

Goldman Sachs is also exploring the use of agentic AI for monitoring trading activity. The bank has significantly invested in AI technology over recent years, especially within its trading and risk management systems. Expanding this work into compliance and trade surveillance shows how major financial firms are leveraging AI to improve oversight. The focus is on creating smarter systems that can analyze complex data streams in real time rather than relying solely on static rules.

This approach aims to strengthen the bank’s ability to detect potential misconduct early while reducing false alarms. By deploying these intelligent agents, Goldman Sachs hopes to better understand the nuances of trading behavior and respond more swiftly to any suspicious activity. The integration of AI into compliance processes reflects the ongoing shift toward more adaptive, data-driven oversight in financial markets.

Overall, these developments demonstrate that leading banks are moving toward smarter, more dynamic systems for monitoring trading activity. The use of agentic AI signals a new era where machines can reason through complex patterns and support human compliance teams in maintaining market integrity. As technology continues to evolve, the future of trade surveillance looks to be more proactive and precise than ever before.

Inspired by

• https://www.artificialintelligence-news.com/news/goldman-sachs-and-deutsche-bank-test-agentic-ai-for-trade-surveillance/

Sources

• unsplash.com
• ai-expo.net
• techexevent.com
• techforge.pub
• techforge.pub