The rapid advancement of generative AI content creation tools equivalent to voice cloning, video and image generation, and chatbots has already given rise to latest types of fraud and financial crime: remember the unlucky employee in Hong Kong who was working there recently tricked into transferring $25 million to fraudsters impersonating his CFO.
But whilst threats multiply – and losses pile up – due to genetic AI, it’s estimated that by 2024, an estimated $485.6 billion can be lost worldwide to fraud Research from Nasdaq — Technologists are working on latest ways to make use of AI and machine learning (ML) to catch and deter those responsible.
Take the startup headquartered in San Francisco humming-birdwhich was launched in 2016 specifically to supply easy-to-use, secure tools for internal criminal investigators and compliance teams at major financial institutions and payment processors (its customers include payments platform Stripe, custom marketplace Etsy, and sports betting platform DraftKings). , FirstBank Puerto Rico and others).
Today, exclusively at VentureBeat, Hummingbird broadcasts a brand new software automation product for purchasers that goals to eliminate the manual, repetitive tasks that occupy investigators and compliance teams at large financial institutions.
“The goal is admittedly bite-sized automation recipes that help investigate financial crimes,” Hummingbird CEO Joe Robinson said in a video call interview with VentureBeat. “And because this work is so regulated, any applications of AI, logic or algorithms generally have to be relatively transparent, highly auditable and process-oriented, with a human within the loop. Those are the important thing features of what we do.”
The dark world of fi crime investigations
Investigating financial crimes may appear to be something out of an episode of Law and Order to those outside the industry, filled with intrigue, drama and excitement.
And that could be it. But Robinson says the core of the work is sifting through massive amounts of monetary transaction data to search out anomalies and determine whether or not they're price looking into, in addition to determining where the cash comes from and where it's going. Is it laundered or is it the results of criminal activity?
For banks and corporations that accept payments, offer payment platforms, account solutions or wallets, the move lately has been to rent teams of dedicated internal investigators to perform this task frequently and to involve external authorities when evidence that a criminal offense has been assembled.
“When you’re employed as a financial crimes investigator, you generally observe the activity of one in all your clients and take a look at to determine, 'Is the activity suspicious or not?'” Robinson explained.
“For example, if I were at a payment card company, I is likely to be 15 to 30 different data sources representing the activity of a specific customer. It's slow, fairly tedious work. Data fragmentation is one in all the essential problems. And later there could also be an audit of why you made certain decisions related to that account and the way you handled them. Therefore, your decision-making process should be transparent. All of these items make this a novel marketplace for applications of AI, machine learning and various algorithms.”
Ideally, Robinson said, the Hummingbird Automations platform will help collect intelligence on fairly routine financial crimes like tax evasion, while freeing human investigators to deal with tougher, less robotically detectable topics like human trafficking.
However, even in the previous case, the platform only presents the collected information to a human official – it’s as much as him to ascertain it for accuracy after which ultimately pass it on as a report back to his superior or external law enforcement, in order that they aren’t within the loop as Expert and controller.
Ready-made “recipes” to automate investigative work while maintaining high accuracy
Hummingbird Automations helps investigators do that with a library of pre-built “recipes,” or workflows, that address key compliance activities at large financial institutions. This includes:
These features aren't nearly saving time; The aim is to enhance the effectiveness and control of monetary crime investigations and make compliance programs more efficient and fewer vulnerable to risk.
“AI might be used to summarize a few of the information available to the investigator and help them interpret what is happening,” Robinson explained. “I believe unlike other markets, AI is admittedly good at identifying trends and anomalies in plenty of data and documentation, so at Hummingbird that’s where we apply it.”
First positive feedback
Robinson told VentureBeat that the automation platform has been within the works for nearly a 12 months and has been in closed beta testing with a small group of about five customers since spring 2023.
Hummingbird provided the next statement from an anonymous Buy Now, Pay Later (BNPL) provider who participated within the beta: “Hummingbird Automations has had a big impact on my team and our work.” We at the moment are capable of provide notifications and automate reports easily, allowing us to enhance visibility and optimize our workflow.”
As financial institutions grapple with the dual challenges of regulatory complexity and the necessity for cost-effective, scalable investigation tools, Automations by Hummingbird delivers latest AI-powered efficiencies while ensuring high levels of transparency, human oversight and compliance reporting.
