In the past year, about 79 percent of financial institutions experienced fraud attempts. As AI-driven attacks outpace traditional defenses, the core challenge is no longer flawless prediction, but the ability to buy time to respond before damage occurs. The authors argue that AI’s primary value in fraud prevention is not merely speed but its capacity to detect uncertainty early and deliberately slow down risky transactions, expanding the window for human judgment. This challenges the conventional "faster is better" mantra. The authors propose a structured, 90-day implementation plan to fight against growing AI-enabled fraud. By deliberately introducing friction and escalating uncertainty earlier, the approach could prevent more losses than perfect, real-time prediction alone.

View the Full Article