Sign in Join
  • HOME
  • Sector Updates
    • Compliance Management
    • Onboarding Verification
    • Financial Crime/Fraud Prevention
    • Identification/Background checks
    • AI & Automation
    • Reporting
    • Transaction Monitoring
    • Communications Compliance
    • Risk Management
    • Cybersecurity/Information Security
  • Legislation Updates
    • KYC
    • AML
    • MiFID II
    • AIFMD
    • Basel III
    • PSD2
    • GDPR
    • Other
  • Features
  • Industry Research
  • Events
    • AML & FinCrime Tech Forum
    • AML & FinCrime Tech Forum USA
    • ESG FinTech Summit
    • Global RegTech Summit – Europe
    • Global RegTech Summit – USA
  • RegTech100
  • Market Maps
    • ESG FinTech
  • Newsletter
  • About Us
Sign in
Welcome!Log into your account
Forgot your password?
Sign up
Welcome!Register for an account
A password will be e-mailed to you.
Password recovery
Recover your password
Search
  • Sign in
  • Contact
  • LOG IN
  • REGISTER
Welcome! Log into your account
Forgot your password?
Register for an account
A password will be e-mailed to you.
Recover your password
RegTech RegTech RegTech Analyst
  • HOME
  • Sector Updates
    • Compliance Management
    • Onboarding Verification
    • Financial Crime/Fraud Prevention
    • Identification/Background checks
    • AI & Automation
    • Reporting
    • Transaction Monitoring
    • Communications Compliance
    • Risk Management
    • Cybersecurity/Information Security
  • Legislation Updates
    • KYC
    • AML
    • MiFID II
    • AIFMD
    • Basel III
    • PSD2
    • GDPR
    • Other
  • Features
  • Industry Research
  • Events
    • AML & FinCrime Tech Forum
    • AML & FinCrime Tech Forum USA
    • ESG FinTech Summit
    • Global RegTech Summit – Europe
    • Global RegTech Summit – USA
  • RegTech100
  • Market Maps
    • ESG FinTech
  • Newsletter
  • About Us
Home RegTech AI & Automation Why AML teams can’t rely on LLMs alone to fight fraud

Why AML teams can’t rely on LLMs alone to fight fraud

October 05, 2026
Why AI alone won't fix financial crime investigations

Financial crime investigators are recruited for their analytical skills. Yet a large share of their working day is often spent on far less valuable tasks: trawling through systems, gathering data, compiling evidence, writing up findings and clearing high volumes of alerts that turn out to be benign.

This problem affects both Transaction Monitoring (TM) and Fraud teams. Analysts may have to move between many platforms before they hold enough information to reach a decision. Rising alert volumes, strict deadlines and more sophisticated criminal tactics are placing more strain on teams that are already stretched.

Artificial intelligence could reshape this operating model. Still, WorkFusion, a UiPath company, argues that simply swapping manual effort for Large Language Models (LLMs) is not the full answer.

Its AI Agent for TM and Fraud Investigations, Isaac, is built on what the company calls a “Goldilocks” principle. This means finding the right balance of AI, deterministic automation and human judgement, and coordinating all three across each investigation.

After a TM alert fires, an analyst may need to pull customer records, review transactions, research counterparties, check case history, compare expected and actual behaviour, map relationships and document everything.

An account takeover (ATO) alert in fraud can require evidence from detection engines, authentication services, card networks, digital identity platforms, behavioural intelligence tools and third-party sources. Isaac is designed to take on much of this “hunter-gatherer” work. In Fraud Alert Review, it connects with existing fraud engines and data sources, consolidates context, analyses transactions, drafts a narrative report and passes the finished work to investigators for human-in-the-loop review.

WorkFusion’s view is that each technology should do what it does best. LLMs add value when unstructured information needs interpreting, such as synthesising historic SAR narratives, reconciling previous dispositions against a new alert or making sense of messy counterparty data.

Transaction arithmetic is a different matter. Aggregation windows, velocity calculations, threshold proximity and peer comparisons should be calculated deterministically and fed to downstream AI as established facts, rather than generated by a model.

The result is that investigators move from gathering and writing to reviewing and deciding. This matters when regulatory deadlines meet unpredictable workloads. Volumes spike, staff take leave and complex cases overrun, which leads to the familiar “day-30 scramble.” AI Agents provide extra capacity by working alerts as soon as they arrive.

Early results look promising. One financial institution using Isaac for first-party fraud and ATO reviews cut manual research time by more than 70%, and its analysts handled two to three times their previous case volume. A regional bank applying Isaac to structuring alerts reported around 10% of alerts auto-grouped, roughly 60% auto-closed and one to three hours saved on cases that needed investigation.

The message for the industry is clear. The future of TM and Fraud is not 100% people, 100% LLMs or 100% rules. It is the right mix of all three.

Read the full Workfusion post here. 

Read the daily RegTech news

Copyright © 2026 RegTech Analyst

Enjoyed the story? 

Subscribe to our weekly RegTech newsletter and get the latest industry news & research

Copyright © 2026 RegTech Analyst

Investors

The following investor(s) were tagged in this article.

  • TAGS
  • Account takeover
  • AI agent
  • AML
  • automation
  • Banking
  • Compliance
  • financial crime.
  • Fintech
  • fraud investigations
  • Isaac
  • LLMs
  • RegTech
  • transaction monitoring
  • UiPath
  • WorkFusion
Previous articleArcher turns AI compliance rules into enforceable code
Next articleWhy SMB verification is failing where volume lives
dwillis

RELATED ARTICLESMORE FROM AUTHOR

VisionX AI’s AEGIS model targets laundering across wallets
Companies

VisionX AI’s AEGIS model targets laundering across wallets

AmEx’s $350m fine exposes the AML blind spot banks ignore
AML

AmEx’s $350m fine exposes the AML blind spot banks ignore

Crypto KYC gaps leave exchanges exposed to regulators
Companies

Crypto KYC gaps leave exchanges exposed to regulators

Latest Analysis

Is regulatory change becoming impossible to manage manually?
Companies

Is regulatory change becoming impossible to manage manually?

November 16, 2026
Can financial institutions keep pace with the changing landscape of AI regulation?
AI & Automation

Can financial institutions keep pace with the changing landscape of AI...

November 09, 2026
Control Room AI faces its toughest test yet: trust
AI & Automation

Control Room AI faces its toughest test yet: trust

November 06, 2026
AI communication and interaction governance in financial services
Companies

AI communication and interaction governance in financial services

November 06, 2026
AML compliance costs plateau as AI savings of $143bn go untapped
AML

AML compliance costs plateau as AI savings of $143bn go untapped

November 06, 2026
  • Term & Conditions
  • Privacy Policy
  • RegTech Analyst’s Editorial mission
  • Contact Us
© Copyright 2026 RegTech Analyst. All Right Reserved

50,000+ RegTech leaders get exclusive industry stories delivered every week

MORE STORIES
Offroad
AI & Automation

Offroad raises $7m to automate AI identity security

June 05, 2026
AI
AI & Automation

Data gaps, not budget, are stalling bank AI plans

September 21, 2026