MuleHunter is based on artificial intelligence and machine learning-based algorithms, and being given free to banks for now, Suvendu Pati, a chief general manager,RBI
Mumbai: At least 15 more banks are set to implement the RBI-developed MuleHunter platform over the next two months to intensify efforts to tackle fraud by identifying accounts used for fraudulent transactions, a senior official said on Friday.
Five lenders, including Canara Bank, Punjab National Bank, Bank of India, Bank of Baroda and AU Small Finance Bank, have already implemented the platform developed by the RBI Innovation Hub, a fully-owned subsidiary of the central bank, while Federal Bank is set to adopt it over the next few days, the official added.
Mule accounts, or accounts created by fraudsters to receive, transfer or launder illicit funds collected from unsuspecting victims, have been a big challenge for the banking system because such accounts, which are generally opened using fraudulently obtained credentials, are used for transferring proceeds of financial frauds.
Usually, there are multiple layers of such accounts from where funds are transferred within a short time to ensure that no clawing back of the funds to the rightful owner is possible.
MuleHunter is based on artificial intelligence and machine learning-based algorithms, and being given free to banks for now, Suvendu Pati, a chief general manager at RBI, said at an ICAI event here.
“We have now implemented across 6 banks. And we are on the course…banks have already got in touch…(to add) another 20 banks or at least 15 banks in the 1.5 or 2 months’ time,” he said.
MuleHunter complements the banks’ systems to tackle such activities, Pati said, claiming that the RBI’s platform is more efficient at detecting such mule accounts.
As against 80 per cent false positives seen in other platforms, MuleHunter has been seen to throw up 90 per cent positive alerts, making it easier for the human behind the software to start acting, he pointed out.
Stating that initial results of the implementation are encouraging, Pati said adding more banks will make the system much more efficient because it will accordingly be able to ‘learn’ and identify patterns deployed.
“The model is still learning, these are early days,” he said, adding that it is used for both savings bank and current accounts. Banks need to devote resources in terms of infrastructure and data scientists while adopting the model.
The platform has identified 90 patterns used by fraudsters operating the mule accounts, including ascertaining that the highest number of such transactions happens between 11 PM and 1 AM because a bulk of the call centres are shut for a variety of reasons, including daily maintenance, leaving no way for the victim to correct the situation.
Pati conceded that as the system is developing and learning, the bad actors will have a field day in the short term, but exuded confidence that over the medium-to-long term, law enforcement will be better.
He also said that the report of a committee on AI set up by the RBI will be coming out soon.
“With quantum computing on the horizon, we also need to be cognizant of the threats posed by the newer technologies and devices, and our response in advance,” the senior RBI official said.