Your fraud program worked. Account history, device signals, and payment data all sit at the platform level, and CE 3.0 rewards exactly that data. Marketplaces win those disputes at scale. Friendly fraud does not attack where you are strong. It has migrated to service-coded disputes: services not rendered, not as described, cancellation and no-show claims. The usual explanation is that these disputes are neglected. New Chargeflow analysis of a full year of Visa and Mastercard outcomes rules that out. Service-coded disputes are contested at higher rates than fraud-coded disputes, not lower. Neglect is not the explanation. The explanation is the evidence itself. Fraud disputes are decided by data that lives in the payment stack. Service disputes are decided by proof of what was actually delivered, which lives in the merchant’s systems. That puts a ceiling on outcomes for any dispute operation working from payment data alone, however well automated. The marketplace is merchant of record and receives the dispute. The merchant delivered the service and holds the proof. Chargeflow’s Dan Moshkovich and Vivek Verma map where that ceiling sits, why it is far lower on one card network than the other, what CE 3.0 did and did not change, and how leading platforms are raising it. Includes a three-question audit attendees can run on their own numbers next week. Findings are presented as relative differences rather than absolute rates.

Detecting fraud and scams is fundamentally a needle in a haystack problem. Most companies use a mix of ML models, rules, expert human analysts, and increasingly AI agents. But each layer of defense comes with tradeoffs around precision, cost, scale, complexity, and regulatory risk. Maximizing the performance of the overall system is an increasingly complex problem. In this presentation, risk practitioners from Coinbase and MouseCat will discuss proven techniques for combining these systems in a way that emphasizes the strengths of each. Attendees will walk away with practical frameworks for how ML models and AI agents can be combined to surface the highest risk activity to human experts, how to operationalize thousands of historic analyst case notes, and much more.

Online harm often translates into challenging offline experiences. This panel examines life beyond the screen - what happens after an incident is over, and individuals move on. We will discuss methods for making amends with impacted individuals, changes to product features that can improve safety and the quality of a user’s experience, and post-mortem best practices.

Increasingly establishing trust and safety doesn’t end at the onboarding stage for most employers. As new avenues for identity and document fraud evolve in the workplace and digital platforms scale, organizations are rethinking how identity verification and risk controls work across the entire user lifecycle. In this discussion, Erin Smalley and Spencer Chee will explore emerging identity trends, evolving fraud patterns you need to be aware of, and practical approaches platforms can use to balance growth, safety, and user experience. Attendees will learn how leading marketplaces are moving beyond point-in-time verification toward more dynamic trust models that help protect platforms while reducing friction for legitimate users.

Fraud is no longer evolving uniformly. Organized fraud operations are increasingly adapting their techniques to specific regions, document ecosystems, and verification environments. The attack affecting a North American marketplace may look completely different from the techniques targeting platforms in Europe, Latin America, or Asia. Drawing on insights from Microblink's latest fraud intelligence research, this session explores how regional fraud patterns are emerging, why AI is accelerating specialization rather than standardization, and what marketplace operators should do to adapt. Attendees will learn: - Why fraud tactics increasingly differ by geography. - How AI is changing the economics of document fraud. - Why static verification models struggle against adaptive attackers. - Practical ways marketplaces can build region-aware fraud defenses without increasing customer friction