The Commonwealth Bank of Australia (CBA) is set to release its artificial intelligence and machine learning tools designed to combat abusive transaction messages. This initiative allows any bank worldwide to access these resources for free.
The AI model is adept at identifying digital payment transactions containing harassing, threatening, or offensive messages within the payment description field. Angela MacMillan, CBA’s Group Customer Advocate, explains, “We developed this technology in response to customers using transaction descriptions to harass or threaten others. Utilizing this model allows us to scan for unusual transactional activity, pinpoint patterns, and identify instances deemed high risk, enabling the bank to investigate and take necessary action.”
MacMillan notes that the model identifies approximately 1,500 high-risk cases annually. This week, CBA is making both the model and source code accessible through its collaboration with H2O.ai on GitHub, the world’s largest platform for hosting source code.
“Sharing our source code and model with banks globally will enhance financial institutions’ visibility into technology-facilitated abuse. This collaborative effort can inform actions banks may take to better protect their customers,” says MacMillan.
This move follows CBA’s earlier pilot with the NSW Police this year, aiming to refer perpetrators of financial abuse to the police, contingent upon customer consent.