Sift

Every fraud scheme begins with an account that should never have been created. For marketplaces and B2B/B2C SaaS platforms with self-serve signup and free trials, registration is the widest open door in the product, and fake account detection is the discipline of closing it without turning away real customers. Account creation fraud is no longer a manual, low-volume nuisance. Advanced AI technology has made it easier than ever for fraudsters, and the data backs that up.

Fraud rarely shows up as a single flagged transaction. It shows up as a pattern, one that’s often invisible to any single business but obvious once you can see across a network. That was the throughline of Sift’s recent webinar unpacking the Q2 2026 Digital Trust Index, featuring Brent Sleeper, Senior Product Marketing Manager, and Jeremy Cannon, Trust and Safety Architect.

The World Cup is the biggest betting event on the calendar. Global wagering on this year’s tournament is already on track to exceed $50 billion, forecast to be the biggest gambling event in history. That also means it’s the most anticipated opportunity for fraud rings. Sports betting platforms know the withdrawal spikes are coming. Every platform running live betting has seen what happens in the minutes after a final whistle.

Online gambling operators face a fraud challenge that’s broader and more operationally demanding than many other digital businesses. The combination of real-money transactions, high-value promotional programs, strict regulatory requirements, and a global player base creates a fraud surface that spans the full player journey, from registration through deposit, gameplay, withdrawal, and beyond.

A study conducted by Public Interest Network found that a staggering 61% of survey respondents have accidentally purchased counterfeit products at least once. This shows just how huge of an issue fraud has become on commerce marketplaces. Fake listings fraud occurs when fraudsters create marketplace listings for products they don’t intend to deliver, deliver in fraudulent or counterfeit form, or use to collect payment before disappearing.

The traditional fraud equation has broken. Fewer attack attempts no longer mean lower losses. Q2 2026 benchmarks reveal attackers have shifted from volume tactics to precision targeting of high-value accounts. Fewer transactions reach your systems, but each one costs more when it lands. Sift’s Fraud Industry Benchmarking Resource shows this shift clearly across all major verticals: fraud is becoming more selective, more efficient, and harder to detect.

Commerce marketplaces face a fraud challenge that differs from single-sided e-commerce platforms, in both structure and complexity. Because two-sided platforms feature both a buyer and a seller side, this creates two distinct attack surfaces for fraudsters to exploit. Marketplace fraud protection addresses threats on both sides of the transaction simultaneously, as well as the platform-level manipulation that coordinated fraud networks use to undermine marketplace integrity.

Multi-accounting fraud occurs when a single person or coordinated fraud ring operates several accounts on the same platform in an attempt to exploit promotional offers, evade enforcement actions, or manipulate platform economics.

Account creation fraud occurs when a fraudster creates an account using false, stolen, or synthesized identity information. Unlike account takeover, which targets existing legitimate accounts, account creation fraud produces what fraud teams often call “born bad accounts.” These are accounts that were never associated with a real, trusted user.

Account takeover (ATO) fraud occurs when a fraudster gains unauthorized access to a legitimate user account, typically to steal stored value, make fraudulent purchases, or exploit the account’s trust status to commit further fraud. For e-commerce platforms, marketplaces, iGaming operators, and SaaS businesses, ATO is one of the highest-impact fraud types in the Fraud Economy because it targets an already-trusted account which bypasses many entry-level controls.