Bank of Ireland cuts customer fraud losses 25% with Visa Featurespace AI

Bank of Ireland has achieved a 25% reduction in customer losses from fraud with Visa's Featurespace, an AI-led pattern recognition technology that can help spot suspicious activity before a payment leaves the bank. Since adopting Featurespace in 2024, the bank has seen a 30% decline in attempted fraud and a 25% fall in customer losses. Featurespace has also helped the bank halve case handling time and achieve an 87.1% drop in false positives, so customers are less likely to be contacted or asked to confirm a payment. There has been an 85% decline in alert volumes, meaning far fewer transactions need to be checked and allowing staff to focus more on higher-risk and potentially suspicious payments. Nine in 10 (90%) customers can confirm genuine transactions without contacting the bank. Nicola Sadlier, head of fraud protection at Bank of Ireland, said: “Combining the latest AI-powered technology through the Featurespace platform with our fraud expertise has transformed our payment fraud detection and prevention.” Payment fraud happens when fraudsters deceive people into sending money or gain access to accounts to make transactions without the customer's permission. The Featurespace platform uses AI and machine learning to help banks and financial institutions identify both existing and new forms of fraud. It uses AI to understand how accounts are typically used, helping the bank identify potentially suspicious behaviour earlier while reducing unnecessary alerts that can delay payments. James Mirfin, SVP and head of risk and security intelligence at Visa, said the work with Bank of Ireland enables the bank to adapt to new and evolving attack patterns without increasing friction for customers. The results have received global recognition from Celent's 2026 Model Risk Manager Award, which the article describes as marking excellent usage of technology by a financial institution and setting a benchmark for future fraud resilience.