The rapid development of generative AI tools—such as voice cloning, video generation, and chatbots—has unfortunately ushered in new forms of fraud and financial crime. A recent example includes a Hong Kong worker tricked into transferring $25 million to fraudsters impersonating his CFO.
As fraud becomes increasingly sophisticated, with an estimated $485.6 billion lost globally in 2024 according to Nasdaq, technologists are innovating with AI and machine learning (ML) to apprehend and prevent these crimes.
Revolutionizing Financial Crime Investigations with AI
San Francisco-based startup Hummingbird, founded in 2016, is dedicated to providing user-friendly and secure tools for in-house investigators and compliance teams at large financial institutions. Notable clients include payment platform Stripe, custom marketplace Etsy, and the sports betting platform DraftKings.
In a media exclusive, Hummingbird announced a new automation software designed to eliminate the manual, repetitive tasks that hinder investigators and compliance teams. “Our goal is to create bite-sized automation recipes that assist with financial crime investigations,” explained CEO Joe Robinson during a video interview. He emphasized the importance of transparency, auditability, and maintaining a human element in AI applications.
Understanding the Financial Crime Landscape
Financial crime investigations might seem thrilling, reminiscent of a “Law and Order” episode, but much of the work involves sifting through extensive financial transaction data to identify anomalies. Investigators must trace funds to determine if they're linked to illegal activities, such as money laundering.
To address these challenges, many financial institutions are employing dedicated in-house investigators who can unravel the complexities of customer activity. “As a financial crime investigator, I analyze customer behavior to assess if it’s suspicious,” Robinson noted. He highlighted the tedious nature of this work, often requiring the examination of multiple data sources, and the need for clear documentation to support decision-making.
Hummingbird's Automations platform aims to streamline investigations of routine financial crimes, like tax evasion, allowing human investigators to focus on more complex cases such as human trafficking. While the platform gathers data, human officers are responsible for validating its accuracy and generating reports for supervisors or law enforcement.
Streamlining Compliance Tasks with Pre-built Workflows
Hummingbird Automations equips investigators with a library of pre-built “recipes” that simplify key compliance tasks, including:
1. Know Your Customer / Business (KYC/B): Monitor customer activities and trigger necessary reviews.
2. Quality Assurance (QA): Schedule reviews to ensure the quality of completed cases.
3. Case Preparation: Expedite investigations by pre-filling case information and findings.
4. Case Monitoring: Automatically alert users about important case updates.
5. Case Management: Automate administrative tasks, including case creation and updates.
6. Activity Digests: Provide scheduled reports on case activities.
These features not only save time but also enhance the effectiveness of financial crime investigations and improve compliance program efficiency.
“AI can help summarize information for investigators and identify trends across large datasets,” Robinson stated, underscoring where Hummingbird’s innovations are focused.
Positive Early Feedback
The Automations platform has been in development for nearly a year and was tested in a closed beta with five customers since spring 2023. A representative from a participating “buy now, pay later” (BNPL) provider shared, “Hummingbird Automations has significantly impacted our team, allowing us to automate notifications and reporting, increase oversight, and streamline our workflow.”
As financial institutions navigate regulatory complexities and demand scalable investigation tools, Hummingbird's Automations offers AI-powered efficiencies while prioritizing transparency and compliance.