How fragmented banking systems became fraud's best weapon, with Feedzai’s Pedro Bizarro

Pedro Bizarro has spent 15 years building the AI systems banks use to make real-time decisions on billions of transactions. As Co-Founder and Chief Science Officer at Feedzai, he has seen what responsible AI requires.In this episode of Banking Reinvented, Pedro joins host Tim Rutten to make the case that safety, fairness, and accuracy in AI are engineering decisions made at the start of the build, not compliance checkboxes added at the end. He walks through FairGBM, Feedzai's open source model that optimizes simultaneously for decision quality and fairness across age and gender, and why the industry was wrong when it told him he had to choose between the two.The conversation covers the asymmetry between banks and fraudsters, why fraud has become one of the top crimes in the world surpassing heroin and cocaine, and why the answer has to involve anonymous cross-institutional data sharing. Pedro also introduces RiskFM, Feedzai's new foundation model purpose-built for financial crime - designed from scratch for millisecond inference at banking scale, not adapted from LLM architecture built for something else entirely.Curious to find out more about AI in banking? Check out our content hub