Survey: Top 3 Challenges Driving AI Adoption in Electronic Communications Surveillance

Introduction

The proliferation of electronic communications (e-comms) platforms, including email, chat, messaging apps, collaboration tools, and mobile channels, has dramatically increased the complexity of monitoring for market abuse, insider trading, and other forms of misconduct.

At the same time, remote and hybrid work models, combined with heightened global regulatory scrutiny, have further strained compliance teams already challenged by fragmented systems and rising message volumes. With millions of electronic communications happening every day, compliance teams can seem like they are fighting a losing battle against sheer volume, and legacy tools may make it worse.

To better understand current operational pain points, market researchers from A-Team Group interviewed senior compliance operations and technology leaders at 16 firms, including investment banks, brokerages, asset managers, and hedge funds. These leaders provided insights on the challenges pushing them to explore AI-driven surveillance capabilities.¹

Where legacy tools are lagging 

Surveillance teams at financial institutions and other regulated organizations are accelerating their adoption of AI-driven electronic communications surveillance to solve three core challenges of legacy architecture:

1. Excessively high false positive rates overload compliance teams. False-positive rates frequently exceed 90-95%, with some firms reporting levels as high as 99%. This may be the most glaring shortcoming of legacy surveillance tech. As daily message ingestion routinely reaches millions of communications (consider that every instant message sent goes toward the total record count), typically fewer than 1% of those generate alerts. However, the sheer scale of e-comms still creates thousands of daily reviews. A Tier-1 global bank reported 7,000 alerts from 5.5 million messages (an alert rate of just 0.1%), but most were false positives, meaning thousands of alerts were likely a waste of time. This volume of noise may force analysts to spend up to 60% of their time on non-issues, driving operational inefficiency and distracting attention from higher-risk behaviors.
   
2. A lack of contextual insight to understand the full narrative behind alerts. Legacy systems are often described as “not sophisticated enough” to understand what surrounds an alert or to connect it to related communications or trade data. In fact, most rely heavily on lexicon-based detection, which respondents characterized as “not very good,” outdated, and inherently inflexible for modern AI or natural-language processing techniques. Alerts from these systems are commonly triggered on standalone keywords without reference to the broader conversion. As a result, benign discussions may be flagged as risky simply because of a phrase match. For example, a discussion about baseball that mentions someone “stealing bases” could trigger an alert that would, of course, end up as a false positive. A lack of contextual enrichment can result in unnecessary alerts, excessive triage efforts, and a more limited ability to identify genuine misconduct.  
3.  Inflexible systems that restrict modernization can lead to roadblocks and bottlenecks. Legacy surveillance platforms can be notoriously difficult to upgrade, customize, or integrate with emerging AI tools. These challenges often force firms to build separate systems or consider full replacements. One French Tier-1 bank reported that its legacy platform was too rigid to support advanced AI, prompting an internal shift to a modular, AI-ready architecture. Manual processes, like the ongoing updating and tuning of lexicons, can add further drag and operational obstacles that can consume compliance resources that could be better spent elsewhere. 
   
   

Solving these challenges with AI

Though there are still some skeptics, AI and large language models (LLMs) are proving to be genuine solutions for surveillance teams. Because of this, financial institutions are increasingly adopting AI and LLMs to help address the limitations of legacy tools and modernize their surveillance capabilities to improve operations, such as:

1. Reducing false positives. The top driver for AI adoption is its ability to dramatically reduce false positive rates. Firms adopting LLM-powered surveillance can see false positive reductions of 30% to 40% or more. One institution cut its alert volume from 900,000 down to 16,000 after LLM tuning — a reduction that transforms the workload and mental burden of a compliance team.
2. Enhancing automation and context. AI-enabled systems can deliver “intelligent filtering” that identifies known benign patterns to help reduce unnecessary alerts. Advanced natural language understanding allows for more accurate recognition of both intent and context. This contextual comprehension can help avoid triggers within harmless phrases like “Let me call you on the phone.” Some firms are even exploring AI-driven bulk-closing of alerts when the system identifies consistent benign patterns across similar communications. Context can help reduce false positives so compliance teams can concentrate on actual risks.
3. Adopting modular, flexible architectures. To continuously improve false positive filtering, firms are investing in modular, adaptable surveillance architectures capable of incorporating new AI advancements without major redevelopment—meaning changes can be integrated more seamlessly without a massive system overhaul. One respondent emphasized designing a new system specifically to benefit from “regular advances in AI.” This modular approach can help compliance and surveillance teams more seamlessly optimize their operations.
   

Download the full report → Electronic Communications Monitoring: Leveraging AI for More Effective Surveillance 

Thoughtfully implemented AI tools can be a win-win

Remote work, new messaging platforms, and rising regulatory expectations have all upped the compliance stakes in regulated industries, while traditional surveillance tools struggle to keep up with the scale, speed, and complexity of today’s communication landscape. AI and LLM-powered solutions are emerging as essential components of a modern compliance strategy, helping to limit false positives, illuminate context, and enable the modularity needed for continuous improvement.

Modernizing systems with the right AI can give firms critical tools needed to help reduce costs and operational pressures, while strengthening the ability to detect real risk before it becomes a regulatory or reputational event. The tipping point isn’t coming. For many firms, it’s already here.

Source: 

1.  A-team Group (2026). "Electronic Communications Monitoring: Leveraging AI for More Effective Surveillance"