Australia’s largest bank, the Commonwealth Bank of Australia (CBA), has taken a significant step in combating online abuse by developing an artificial intelligence (AI) tool to block abusive messages in digital transactions. The AI model is designed to identify and filter out digital payment transactions that contain harassing, threatening, or offensive messages. Since its implementation in 2020, the automatic filter has successfully blocked close to one million transactions.
The need for such technology arose when the bank noticed that some customers were using transaction descriptions to harass or threaten others, an act known as financial abuse. Angela MacMillan, the CBA customer advocate, explained that the AI model scans for unusual transactional activity, identifies patterns, and highlights instances that are considered high risk. This enables the bank to investigate and take appropriate action. In an unprecedented move, the CBA has made its source code and model freely available to banks worldwide. This aims to enhance the visibility of technology-facilitated abuse and empower financial institutions to better protect their customers.
Collaborating with global AI company H2O.ai, the CBA developed this groundbreaking AI model. As part of the pilot program, the bank’s Next Chapter team intervenes if a customer is repeatedly subjected to abuse through transaction descriptions. With the customer’s consent, the CBA reports the abuse to the New South Wales (NSW) Police on their behalf. Plans are underway to establish a streamlined process that allows the bank to report abuse with the consent of the victim-survivor, further streamlining the reporting mechanism.
Since June 2020, the CBA has actively monitored and blocked abusive transactions, with hundreds of thousands being intercepted each year by the automatic filter. Human review, combined with AI assistance, forms an integral part of this process. Alongside this development, the CBA has reported a 6% increase in net cash earnings compared to the previous financial year, reaching $10.1 billion (US$6.47 billion). The rise in interest rates contributed to a 2.07% jump in net interest margin (NIM), although intensified competition partially counteracted this growth.
While occupancy and equipment expenses decreased by 3%, thanks to optimization efforts in the bank’s digital, branch, and ATM network and the exit from commercial office space, information technology expenses rose by 3%. This increase can be attributed to inflation, software licensing, and infrastructure costs. Additionally, staff expenses increased by 9% due to wage inflation, the hiring of more full-time equivalent staff, and increased annual leave. The bank’s productivity initiatives managed to offset other expenses by 2%, although the lifting of COVID-19 restrictions led to higher travel costs.
In September, the CBA made headlines by tripling cash deposit fees for business owners. Nevertheless, the bank emphasized its commitment to supporting customers who prefer cash payments and assured them of continued services.
The CBA’s pioneering development of an AI tool to block abusive messages in digital transactions sets a significant precedent. By making its source code and model freely accessible to global banks, the CBA aims to create better visibility of technology-facilitated abuse and foster collaboration in protecting customers. With the combination of AI and human intervention, financial institutions can proactively combat financial abuse and ensure safer digital transaction experiences for all.
