Using Privacy Preserving Analysis to Tackle Financial Crime

Privacy enhancing technologies, or PETs, may provide a “similar or, even, identical utility” compared to analysis of the raw data in some use-cases, says an FFIS paper.

The FFIS (Future of Financial Intelligence Sharing) research programme has published a new discussion paper on the use of ‘privacy preserving analysis’ to tackle financial crime.

Privacy preserving analysis is used to address limitations on the nature and extent of information-sharing that is usually possible due to legal, regulatory and technical constraints.

It relies on ‘privacy enhancing technologies’, or PETs, which are specialist cryptographical capabilities. “Such privacy-preserving analysis allows for computations to take place on underlying data, without the data owner necessarily divulging that underlying data,” the paper says.

While access to information in its raw unencrypted form would provide a maximum level of utility, PETs increase the level of privacy protection – though reducing the range of query capabilities.

For specific use-cases where all data queries are not required and access to all underlying data is not required, PETs may provide a” similar or, even, identical utility” compared to analysis of the raw data, the paper says.

It highlights the FCA’s TechSprint last July, and a follow-up workshop in September 2019, which focused on the use of PETs and data analytics to develop solutions to facilitate information sharing in the context of AML and financial crime.

The FCA is now piloting a ‘digital sandbox’ to support access to high-quality data assets including synthetic or anonymised data sets to enable testing and validation of technology solutions, including PET solutions, to financial crime prevention use-cases.

The FFIS paper compiles available case studies on the use of privacy preserving analytics for AML or financial intelligence objectives, to further build awareness of the technology and promote its consideration in policy and supervisory initiatives.

One of the case studies highlights the work of AUSTRAC in this area, after it was granted AUD 28.4 million over 4 years in the 2019/20 Australian federal budget to expand the Fintel Alliance, including to develop operational capabilities using privacy preserving techniques.

AUSTRAC’s algorithm is designed to flag suspicious networks from domestic retail account and transaction data, while protecting the privacy of all data and using only six data fields. The project is ongoing and AUSTRAC is currently exiting the “discovery” phase of the project.

“A key lesson identified by AUSTRAC during the discovery phase is that there is not a ‘one size fits all’ algorithm for any given use case,” the paper says. “AUSTRAC expects the algorithm to require tailoring for each specific use case/operation.”

Together, the case studies provide an indication of the pace of development of privacy preserving analysis in the field of AML and financial investigations.

The next decade will see significant growth in the standards framework for underlying PET technologies, the efficiency and effectiveness of using these techniques in practice, and greater expectations to use privacy preserving analysis, the paper says.

The full paper is available here.

FFIS is a research and international knowledge exchange programme specifically focused on the role of public–private information-sharing to improve AML/CTF outcomes. The programme is a research partnership between the RUSI Centre for Financial Crime & Security Studies and NJM Research, supported by Verafin, Oliver Wyman, Western Union, Refinitiv and The SWIFT Institute.

To Top