At the 6th Financial Innovation Forum, experts highlight how AI tools can spot suspicious activity earlier in the scam process, yet emphasise that final decisions should still rely on human judgment to prevent scams effectively.
Artificial intelligence is increasingly being pitched as a way to catch fraud and scams earlier, but the harder problem is deciding when a payment that looks legitimate is actually the product of manipulation. At a panel at the 6th Financial Innovation Forum, Vivox AI founder and chief executive Tim Khamzin argued that banks and payment firms need systems that can detect suspicious patterns before money moves, while still leaving the final judgement to people.
Revolut’s Natalia Gburzyńska, speaking in a personal capacity, described scammers as “emotional illusionists” who exploit trust, fear and distraction. She said urgency is the enemy of proper scrutiny, because a warning that arrives after a customer has already committed emotionally may be too late to stop the transfer. Vivox AI said even a short delay can disrupt a fraudster’s momentum, particularly in romance and investment scams where a conversation with trained staff may be needed to break the script.
Pallavi Kapale, senior financial crime officer in the financial intelligence unit at Bank of China UK, gave an example of a customer in Spain who sent money to what she believed was a military boyfriend, only to lose £65,000 over six months. She noted that Confirmation of Payee, the UK system that checks whether an account name matches the sort code and account number, would not have exposed that lie because a matching name does not prove the relationship is genuine. John Sudbury, threat intelligence lead at Wise, said firms should map the fraud “kill chain” and look for behavioural changes earlier, such as unusual app activity or shifts in voice cadence, rather than waiting for the final payment request.
Anne Markey, managing director at Alvarez & Marsal, called for a layered approach that combines authentication, biometrics, liveness checks, IP data and controls against injection attacks, pointing to her own work on machine learning for anomaly detection at UBS more than a decade ago. The panel drew a clear line on automation: AI can support decision-making, but it should not replace human judgement. Vivox AI’s broader message was that success will belong to tools that close the gap between the first warning sign and a meaningful intervention, while still producing decisions that can withstand audit scrutiny and make sense to a customer who has been misled.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
