Resistant AI has been
named ‘Digital Crime Fighter of the Year' for winning the inaugural hackathon
challenge at the recent ACAMS Hollywood conference in Florida last week. The
hackathon participants were given a large and complex synthetic dataset with
more than a million rows that included customer information, wires, credit card
and ATM activity and tasked to uncover potential human trafficking activity.
The Resistant AI solution won by quickly assessing data and
uncovering anomalous patterns to detect behaviors indicative of human
trafficking offences. While its almost impossible to measure the scope and
scale of human trafficking, according to the International Labour Organisation
(ILO), there are over 40 million victims of modern slavery worldwide. The
prevalence of human trafficking was highlighted further in recent months, as
reports circulate of traffickers targeting vulnerable women and children
fleeing the crisis in Ukraine.
The hackathon, sponsored by PwC, represents an urgent need
to use intelligence in the fight against financial crimes that reach far and
wide around the world. Judges for the event were Jim Candelmo, chief bank
secrecy act and AML sanctions officer at PNC; Melissa Strait, chief compliance
officer at Coinbase; and Scott Nathan, head of AML transaction monitoring at
Citi.
"The teams participating were nothing short of incredible,"
said Vikas Agarwa, partner at PwC, Financial Crime Unit Leader. "Each group
demonstrated multiple unique approaches and attacks -- each of them focused on
how, through purpose and innovation, they quickly identified some or all
elements of our hidden human trafficking data."
"Resistant AI analyzes the hidden relationships between
identities and transactions to draw a better decision boundary between
legitimate and criminal activities," said Martin Rehak, CEO of Resistant AI.
"Our team took a data-driven approach to the hackathon exercise, using the
anomaly detection tools in our AI-powered identity forensics solution."
Resistant AI's advanced analytics solution helped the
hackathon team more effectively pinpoint anomalous and suspicious behaviors and
also uncover previously undetected patterns. Through a variety of machine
learning techniques, the team produced results that uncovered the chain of
human trafficking activity.
The challenge is that many organizations are still relying
on traditional scenarios or rule-based systems to find and root out suspicious
behavior. While these systems act as a useful baseline, they only look for
scenarios that are known as suspicious, and miss the unknown patterns of
behavior. Even when new patterns are detected, there is a delay in defining
rules that will detect those patterns in the future. The volumes of alerts and
false positive rates of traditional systems also make it extremely difficult
for investigators to identify true risk amid the noise.
"Congrats to the winner, Resistant AI, that showcased the
power of innovation with tangible, data driven approaches to uncovering
unusual, high-risk behaviors indicative of human trafficking across large
complex data sets," said Scott Nathan, head of AML Transaction Monitoring at
Citi. "Gone are the days of multi-year, linear ‘transaction monitoring' system
implementations. Today it's all about ‘customer interactions' across
multi-dimensional data constructs (internal and external), agile development,
and the confidence to fail-fast while processing petabytes of data at scale."