Tackling financial crime, and ensuring compliance with anti-money laundering (AML) and combating the financing of terrorism (CFT) regulations remains one of the biggest overall challenges for the financial services community.
This year, this has been particularly apparent, as the quick shift to enable a remote working environment in the wake of the pandemic put additional pressure on the fraud defences financial institutions have spent time and money investing in.
The ever more connected and instant digital world also poses questions for how banks and financial institutions can ensure they meet their compliance requirements in real-time. In one of two risk management sessions on day 2 of Swift's Community Update - Focus on Europe virtual event, Swift's vision for compliance in an instant and frictionless cross-border payments world was put under the microscope.
Areas to tackle
Looking at cross-border payments, Stefano Favale, global head of GTB at Intesa Sanpaolo, identified KYC and data protection as two of the main compliance pain points to be solved.
"KYC is an issue particularly for corporate treasurers, where it is considered the most impactful activity in most cases," Favale noted. "The KYC space grew a lot over the past year in terms of activities and complexity. It is something corporates and banks have to tackle. Most corporate RFPs these days consider KYC on the bank side as a key point. Corporates might decide to reduce their number of banks, or not to switch premium business from one bank to another because of KYC constraints."
Swift has moved to address this area of the corporate-bank relationship with its KYC Registry. The solution, originally for banks, was opened to Swift’s corporate customers in December 2019, following a testing phase with 18 corporate groups including BMW, Spotify and Unilever supported by 16 global banks representing over 7,000 corporate to bank relationships on Swift.
Another key financial crime topic this year has been the increase in fraud attempts that have tried to take advantage of the Covid-19 pandemic. In a world where retail cross-border payments are becoming more common, participants operating in this business area need to implement the right financial crime compliance procedures in order to secure the payment chain for consumers and their counterparts.
"What is emerging now, particularly on the corporate side, is that instant prevalidation is definitely a value added service," Favale said. "This will also allow the banks to perform some compliance screening."
Swift on compliance
With Swift's platform transformation and move towards transaction management set to enable instant and frictionless transactions, the industry cooperative is also hugely invested in ensuring that it gets the financial crime compliance side correct. As Robert Schneider, head of financial crime compliance & data, EMEA at Swift explained, a lot of existing friction is caused by compliance and fraud detection checks, something that the migration to ISO 20022 will seek to smooth out.
"With transaction management, the 'immutable copy' of full transaction data is maintained at the centre, ensuring end-to-end transaction integrity. Intermediaries can still add information on fees or on additional intermediaries, but they cannot remove anything related to the originator of the message or the beneficiary, for example. Also, the full transaction data will be available across the chain to all participants. This means that, no matter which format or channel you are using to interact with the platform, you can benefit from the central copy of the data and can screen against the full data set."
Swift is future proofing all of its compliance services, including those that sit within the transaction flow, and those that provide data and analytics for compliance and fraud detection. As it looks forward, Schneider said Swift is exploring real time analytics and machine learning capabilities to make alert management even more efficient.
"The ideal scenario is to have real time analytics that not only detect fraud, but can actually prevent fraud in the first place," he explained. "Again, this will enable us to remove frictions and enable instant payments."
The approach Swift is taking to its enhanced fraud and financial crime compliance tools is one of ongoing evolution, Schneider emphasised, in order to bring the whole community along with it.
"By doing this, existing users will benefit from the enrichment of the existing tools," Schneider noted. "We also want to evolve and support additional customer segments, meaning we want to serve the whole Swift community with those solutions going forward."
Customer insights on cutting edge tools
The financial crime compliance session closed with a panel discussion around how the financial services sector can achieve safe, compliant and faster cross-border payments. The use of new and evolving technologies in this field dominated the discussion.
Accuracy and integrity of data is key if tools such as AI and machine learning are to be used successfully in financial crime compliance. Salome R. Birgisdóttir, head of back office at the Central Bank of Iceland, shared an example from Iceland, which has a national registry with every person assigned a 10-digit ID number.
"My 10 digit number identifies me as Salome R. Birgisdóttir, and wherever I go this can be used to produce my name completely, with no abbreviations and no spelling mistakes - the sorts of things that can make a lot of noise in screening and compliance," she said. "So far this has only been possible in a small community like ours because of the data issues. But I see huge possibilities in machine learning and artificial intelligence here."
When it comes to AI, particularly in compliance assessment, the journey is one of evolution, not revolution, as intelligent tools learn on the job from the data they are provided with.
"Most of those who are using AI models are working with supervised models, essentially models with teachers," commented Larisa Zalomikhina, head of compliance at Sberbank. "They are delivering scoring and assessing areas that were previously assessed by humans. You can't create a model in absence of data - it is all about looking at your history and what decisions were made previously. That is why it is an evolution."
Bertrand Godin, head of operations & correspondent banking at iBanFirst, shared an interesting perspective on why fintechs and non-banks may struggle to implement AI in transaction management fraud compliance compared to banks.
"Many fintechs and non banks have fewer data sets to start with than banks do, which makes it quite hard to be able to start defining these rules for AI and work on them," he said. "Also, many companies like us are very quickly changing in terms of client type and payments type targets. The model that was built could just change really fast, so the data of last year might not be relevant for the data of this year."