Fraud Prevention in the Digital Age from KYC to Chargebacks - with Matt DeLauro of SEON
🎯 Summary
Podcast Episode Summary: Fraud Prevention in the Digital Age from KYC to Chargebacks - with Matt DeLauro of SEON
This 24-minute episode of the AI and Business Podcast, featuring Matt DeLauro, President at Cian Technologies, provided a comprehensive overview of the evolving challenges and advanced solutions in digital fraud prevention across various industries, emphasizing the need for agility and early intervention.
1. Focus Area
The discussion centered on advanced fraud prevention workflows in the digital age, moving beyond traditional, siloed checks. Key themes included the limitations of legacy fraud systems, the necessity of data aggregation, the impact of “Fraud as a Service” on sophisticated attacks, and the strategic implementation of “shift left” methodologies. A significant portion was dedicated to tackling the growing problem of chargebacks, particularly those stemming from first-party fraud, and leveraging AI for dynamic defense.
2. Key Technical Insights
- Shift Left and Implicit Data Capture: The industry is moving toward introducing frictionless, observable capabilities much earlier in the customer journey (shift left). This involves capturing rich, implicit data—such as IP information and behavioral biometrics—in the background rather than relying solely on hard, friction-inducing Know Your Customer (KYC) checks later on.
- Dynamic Friction Implementation: The most effective fraud teams implement dynamic friction, profiling visitors pre-KYC to simultaneously remove friction for high-value customers (improving conversion) and increase barriers for potential fraudsters.
- LLMs for Dynamic Argumentation in Chargebacks: Large Language Models (LLMs) are drastically lowering the cost and effort required to construct dynamic arguments to fight chargebacks. By feeding LLMs rich data (device fingerprinting, location data, session history), they can construct powerful, narrative-based defenses that were previously manual and expensive to create.
3. Business/Investment Angle
- Agility Over Legacy Systems: The primary business challenge is that digital experiences (new products, markets) iterate faster than legacy fraud systems can adapt, creating significant pressure on fraud teams to maintain assurance levels.
- Democratization of Sophisticated Fraud Tools: The rise of “Fraud as a Service” means sophisticated attack tools (like those leveraging generative AI for profile creation) are accessible to smaller fraud rings, forcing legitimate businesses to adopt equally advanced, multi-perspective defense mechanisms.
- Urgency in Retail vs. Financial Services: Retailers face a higher sense of urgency regarding fraud prevention because rising chargeback rates can immediately impact revenue (e.g., losing payment methods), whereas financial institutions often face regulatory penalties that may materialize over a longer timeline.
4. Notable Companies/People
- Matt DeLauro (President, Cian Technologies): The expert guest, providing insights from Cian’s work in providing AI/ML-driven fraud prevention across the customer journey.
- Cian Technologies: The sponsoring company, highlighted for its solutions that aggregate rich session data and leverage AI for fraud mitigation.
- Tacobas (Mentioned Example): Used as an example of a business selling soft goods where the marginal utility of winning a chargeback is high, even if the average transaction value is low, necessitating automated defense.
5. Future Implications
The industry is moving toward highly unified, data-driven fraud views that integrate data silos (e.g., AML, payments fraud, KYC teams) within large institutions. The future of defense lies in proactive, background monitoring using behavioral biometrics and device intelligence, coupled with AI-driven automation for reactive processes like chargeback defense, ensuring security without sacrificing customer conversion rates.
6. Target Audience
This episode is highly valuable for Fraud Operations Leaders, Risk Management Executives, Data Science Teams involved in anomaly detection, and Technology Strategists in the Fintech, E-commerce, and Retail sectors who are navigating the transition from legacy compliance checks to modern, AI-augmented defense postures.
🏢 Companies Mentioned
đź’¬ Key Insights
"I think what's happened with LLMs is it's now made a lot of the cost structure around making dynamic arguments like it's virtually like no cost around it whatsoever. So you don't really need the human cost around making dynamic arguments."
"That first party stuff is really tough not to crack. And that's what's growing in order of magnitude faster than the third party stuff."
"I think the biggest challenge over the last couple of years has been the growth in first party fraud resulting in a charge back. It's particularly dangerous in the case of returns fraud because then you could be out the revenue and the product. A double whammy from a loss perspective..."
"The average card value might be $2,000, but then if it's $20, you really have to find a way to dynamically fight those charge backs that has as little human intervention as possible. Because it's not worth it. This has been $25 to fight a $20 charge back."
"That big shift left away from just doing a hard KYC check. I'd say as soon as somebody's on your page, what kind of IP information are you getting on them? Not just to prevent bots, but also to try and really understand what the behavioral biometrics look like for a good customer who's visiting the site the right way..."
"there's not a relative technology, just as an example from the fraud side that we have for merchants and retailers that would match necessarily say a corrupted version of chat GPT. That can create a zillion profiles or things like that or a zillion emails, all inquiring about the same kind of product that those forms of fishing and spam fraud."