Commonwealth Bank of Australia (CBA) has announced that it will make its artificial intelligence and machine learning tools, designed to combat abusive transaction messages, freely available to banks globally.
The AI model is specifically engineered to detect digital payment transactions that contain harassing, threatening, or offensive messages within the payment description fields. Angela MacMillan, CBA’s Group Customer Advocate, emphasized the need for this technology, stating, “We developed this solution after observing that some customers exploited transaction descriptions to intimidate or threaten others.”
Utilizing this model allows the bank to monitor unusual transactional behaviors, identify patterns associated with high-risk activities, and investigate instances that raise alarms. According to MacMillan, the model successfully identifies approximately 1,500 high-risk cases each year.
This innovative tool and its accompanying source code will be released this week through CBA’s collaboration with H2O.ai on GitHub, the premier platform for hosting source code. MacMillan added, “By sharing our source code and model with any financial institution around the world, we aim to enhance their awareness of technology-driven abuse, enabling them to take appropriate measures to safeguard their customers.”
This initiative follows a successful pilot program earlier this year in collaboration with the NSW Police, allowing the bank to refer cases of financial abuse to law enforcement with the customer’s consent.