Banks Have Struggled to Implement AI Trade Surveillance
Data fragmentation and aging infrastructure remain the primary barriers to deploying AI-driven monitoring tools in finance.
Updated on Sept. 21, 2026 in Artificial Intelligence
Live Poll
Do you trust that businesses can successfully implement new AI tools with their existing data quality?
While 89% of banks seek to implement AI-enhanced trade surveillance, only 11% have successfully deployed these systems. Poor data quality and fragmented capture processes prevent most institutions from scaling AI tools effectively.
Why it matters
The gap between intent and implementation highlights how technical debt and data silos in banking prevent the adoption of modern risk-monitoring software. Current infrastructure often forces reliance on manual alerts rather than automated, AI-driven oversight.
A 93% majority of surveyed banks identify false positives as a meaningful operational drag, yet 48% of firms still operate with completely unlinked surveillance controls. No institution currently merges trade and communications data before the alert generation stage.
The players
1LoD
An industry firm focused on operational risk management that conducts benchmarking research for financial institutions.
The details
Effective AI surveillance requires clean, normalized data streams, but 71% of banks cite fragmented information and lack of standardized formats as major hindrances. Four in five institutions rely on legacy, outdated platforms that struggle to ingest data before the alert stage, causing persistent inaccuracies. This creates a technical bottleneck where modern AI models lack the cohesive input necessary to reduce the high rate of false positives.
Timeline
1LoD conducted the Surveillance Benchmarking Survey in 2026.
In 2024, 59% of banks viewed regulatory risk as a primary limit to implementing new surveillance technologies.
The Tech Race
The 2026 1LoD Surveillance Benchmarking Survey marks a departure from the common assumption that regulatory caution is the primary hurdle for bank AI adoption. Instead, the results quantify a deeper infrastructure crisis where data fragmentation prevents banks from competing with agile, AI-native fintech platforms.
Banking analysts should expect continued reliance on manual review processes while firms prioritize the expensive, multi-year task of consolidating legacy data silos. Until institutions successfully integrate trade and communications data, the operational drag of false positives will continue to limit analyst productivity.
The takeaway
The primary barrier to AI in finance is not model capability, but the persistent inability of legacy banks to unify disparate data streams. Watch for future benchmarks regarding data-linkage maturity to confirm if firms are prioritizing infrastructure upgrades over model experimentation.
Further reading
For broader trends in enterprise adoption, visit Artificial Intelligence.
Live Poll
Do you trust that businesses can successfully implement new AI tools with their existing data quality?
