Design a Feature for User-Reported Fraud/Scams

To design a fraud reporting system for Stripe, I would prioritize minimizing latency while maintaining high accuracy. First, I'd establish clear success metrics: reducing time-to-block for verified scams to under five seconds and keeping false positives below one percent to protect legitimate merchant revenue. The system would start with a unified reporting interface accessible via API and dashboard, allowing users to submit transaction IDs, screenshots, and contextual notes. This data feeds into a real-time risk engine that cross-references historical patterns, IP reputation, and behavioral biometrics. If confidence scores exceed a dynamic threshold, the system automatically flags the transaction and freezes funds pending review. Crucially, we need a human-in-the-loop escalation path for edge cases where AI uncertainty is high. We would implement a rapid-response team workflow that prioritizes reports based on severity and potential loss magnitude. To prevent abuse, we'd introduce rate limiting and reputation scoring for reporters themselves. Finally, the system must be transparent. Affected users receive immediate status updates, and we'd publish monthly transparency reports on blocked attempts. This approach aligns with Stripe's mission to increase GDP by building trusted infrastructure, ensuring that security measures enhance rather than hinder the payment flow.