October 14, 2025
Introducing Ti2: The next generation of Plaid Protect’s Trust Index
Five months ago, we launched the Trust Index (Ti1)—Plaid Protect’s machine learning-powered fraud score. Built on the strength of Plaid’s bank data network, it gave fraud teams a new kind of signal, one based on data that fraudsters can’t fake.
Today, we’re announcing the next version: Ti2.
From the start, the Trust Index was built for rapid iteration, because fraud is constantly evolving, and staying ahead means moving faster than the attackers. With twice the training data and a sharper view of user behavior, Ti2 already catches 30% more fraud, and spots patterns that traditional tools still miss. From patterns in transaction histories to network-level signals, Ti2 adds depth and context to every fraud score, starting from the moment a user touches your app or service.
What’s new in Ti2?
Ti2 builds on the foundation of the original Trust Index with two major upgrades: bank account transaction history and user graph insights. These help detect harder-to-catch fraud like synthetic identities, account takeovers, and coordinated attacks across apps.
Bank account transaction history
Ti2 now analyzes transaction history to better understand how accounts behave historically. This provides stronger signals about whether an account is acting like a real user, or a fraudster.
For example:
If a user typically pays a phone bill once a month but suddenly sends multiple peer-to-peer transfers to never before seen recipients at 3 a.m., that could be an ATO.
If funds arrive from a crypto or fintech app and are immediately sent out to several unrelated accounts, we may have found a money mule.
If a user quickly cycles money through high-risk merchants or rarely seen categories, that might signal first-party fraud.
In one real case, Ti2 flagged an account receiving repeated payments from popular trading apps and peer-to-peer money sharing apps, followed by ATM withdrawals and a series of $499.99 purchases at Best Buy, the exact price of a PlayStation 5, a common item used in fraud rings. Without Ti2’s transaction insights, this pattern wouldn’t have stood out. With it, the signal was clear.
User graph insights
Ti2 also now includes live, network-wide graph analysis. From these patterns of shared risk across users, devices, accounts, and apps we can surface fraud rings that are invisible when looking at one app in isolation.
Ti2 detects both the obvious patterns (like dense clusters of reused accounts) and the subtle ones (like wide webs of lightly reused bank accounts). These graph-based features are now built into the model and evaluated in real time. In one example, a single device attempted to sign up for dozens of different apps, all in a tight time window. Each time, the user was behind a different VPN or proxy. On their own, these users looked unrelated. But zooming out, Ti2 recognized them as a coordinated, identity-cycling operation, using the same device and jumping IPs to avoid detection.
The proof is in the performance
With 2x the training data behind it, spanning more fraud types and more diverse user behavior, Ti2 is already delivering impressive results in retro analysis:
For a large digital lender, Ti2 would have detected 40% of first party fraud that their existing solution couldn’t by stepping up just 5% of their users.
For a leading cash advance service, Ti2 would have stepped up 1 in 10 users joining their platform to catch 43% more fraud.
For a major crypto firm, the Trust Index could have prevented 50% of their fraud losses.
Ti2 evolves the Trust Index into a more powerful, data-rich engine for fraud detection, delivering broader visibility and sharper accuracy.
Ti2 is just the beginning
While Ti2 delivers a huge leap against fraud today, it’s also the start of something much bigger.
"We're still in the really early innings of the Trust Index, and there's a long list of high-signal fraud data that we want to bring into the score—we've just started working through it. The future's kind of awesome."
— Alain Meier, Head of Fraud at Plaid
As more intelligence sources, from graph neural networks to AI agent analysis, are integrated into the model, the Trust Index will continue evolving. You can expect even sharper scores, more proactive defenses, and smarter decisions throughout the user journey.
Ready to catch more fraud?
Whether you’re battling ATOs, synthetic identities, or first-party fraud, Ti2 gives you unmatched visibility, and the power to act with confidence.
If you're interested in joining the Ti2 beta, reach out to your Plaid account manager. Or, sign up for the waitlist to be notified when Plaid Protect becomes more widely available.