Case study

Absa finds an ally in AI for crime detection, reducing false positives by 77%


Download

 
The global tier 1 bank incorporated AI into their financial crime management, identifying new risks and increasing their hit rate

The challenge

Absa is one of the largest banks in Africa, working with a global client base. They partner with SymphonyAI for anti-financial crime compliance solutions, including transaction monitoring, KYC/CDD, and sanctions screening.

The finance industry and its regulations are becoming increasingly complex, so a dynamic, modern, and forward-looking approach is required. Alongside this, there is an urgent need to combat the threat of rising financial crime, with criminals using AI tools to test and threaten bank processes.

Keen to continue leading the industry with an innovative approach to anti-financial crime, the bank wished to test the impact of AI on their teams’ productivity and risk reduction, specifically using the technology to test against their current transaction monitoring rules-based solution. If successful, it would provide enhanced alert quality and risk detection, valuable use cases across the bank, and cement Absa as an industry pioneer.

The SymphonyAI Solution

With AI exploding in popularity, productivity and efficiency claims of software are everywhere. Absa was interested in using the technology but wanted to see the results with their own dataset. The SymphonyAI response was a series of proof of concepts (POCs) designed to prove the likely positive impact of using AI technology across the company, from efficiency and effectiveness of alerts through to potential cost savings.

SymphonyAI worked with Absa’s masked data, developing new AI models and software features. Proposed advancements included improving and speeding up new risk detection beyond the bank’s current achievements alongside improving alert quality – reducing false positive alerts while continuing to identify all known suspicious activity.

Alongside this, SymphonyAI combined five different language learning models that crossed the full spectrum of manual to supervised analysis, while also developing new features to further enhance the project.

The result

The outcome far exceeded expectations of the proof-of-concept project, persuading Absa to further collaborate with SymphonyAI to fully implement the solution and roll out to other jurisdictions across the continent.

Highlight results included:

  • SymphonyAI reduced false positive alerts by 77% while also capturing all suspicious activity found using their current transaction monitoring system.
  • Many potential new risks were discovered, which were filtered down to the 200 highest scoring risks based on transaction amount, average score, and maximum score.
  • The bank confirmed 21 new risks were identified quickly versus a slower manual adjustment of rules or manual reporting methods. Alongside this, the new risk identification hit rate of 10.5% was significantly more effective than using rules alone.
  • The project was so successful that the partnership was shortlisted for an International Compliance Association award.

The Future

The outcome of this project has seen three more AI proof of concepts planned in areas such as entity resolution and watchlist management, showing Absa’s commitment to continuing the AI journey, adapting to new market demands, and being an innovator in the fight against financial crime.

Latest Insights

regtech company of the year
 
Case study

SymphonyAI wins RegTech Company of the Year at the FinTech Awards London 2024

Financial Services Square Icon Svg
 
Case study

Cecabank partners with SymphonyAI to revolutionize its retail and wholesale banking security

Financial Services Square Icon Svg
SymphonyAI Financial Services wins ICA Compliance Award
 
Case study

SymphonyAI Financial Services wins ICA Compliance Award

Financial Services Square Icon Svg