OmniTrust Capabilities
Proof's OmniTrust capabilities feature the Defend Risk Model, a machine-learning fraud detection system that integrates extensive network data, biometric and credential matching, deepfake and liveliness detection, and payment network signals to comprehensively identify and prevent impersonation and fraudulent authorizations, while also introducing portable identity for AI agents to prove their authorization.
New
Proof gives AI agents a way to prove what they're authorized to do
Introducing portable identity
Defend Risk Model
Fraud has nowhere to hide. Defend Risk Model sees it all.
Defend is powered by Defend Risk Model, the first ML fraud model purpose-built to flag impersonation for any type of authorization.

Data from our network provides the fraud intelligence to protect every authorization
Defend Risk Model builds a comprehensive risk profile by integrating traditional identity signals with data from our extensive network. This approach provides a complete understanding of an individual and the things they agree to.
Deepfake detection
Proof has secured millions of authorizations over face-to-face video meetings with over 600K hours of video. Defend Risk Model has been trained to spot deepfakes on video.
Credential and biometric matching
Defend Risk Model identifies and flags inconsistencies such as a single individual registering as multiple users or a biometric scan not matching the registered user.
Liveliness detection
The machine learning model detects physical presence to protect against replay attacks or AI impersonation.
Payment networks signals
Defend Risk Model incorporates signals from payment networks to assess the risk associated with prior fraud indicators tied to a credit card.
Synthetic identity detection
The fraud model flags fake identities by detecting discrepancies in personal records or behavioral patterns.
Human-in-the-loop feedback
Thousands of agents help identify fraud and train the model when they report a suspicious user. Our network of agents provide a human in the loop.
IP and location data signals
By comparing a user's current IP, geolocation, and device context with historical patterns, the fraud model flags suspicious activity like impossible travel and the use of VPNs.
Device fingerprinting
Device intelligence helps catch coordinated fraud patterns like if one mobile phone is associated with multiple identities.
Transactional data
Defend Risk Model understands the nature of the transaction and incorporates that into the model. A suspicious phone number can carry different risks on a $100K retirement withdrawal vs. a mortgage application.
Protect your business with Proof
Verify your customer’s identity and start building a trusted relationship in minutes.