FiVerity

Machine learning detects synthetic identities that have no fraud history, and pattern matching lets institutions compare cases without sharing personal data. Rules based systems, which the company was founded to beat, are what removal would leave.

Each institution's own fraud analysts work alongside the machine learning, joint investigations run through structured packages with role based access, and suspicious activity reports are prefilled for the institution to review and file. Confidence thresholds on network alerts, and what an analyst should verify first, are not published.

Detection is framed against the alternative, catching more than half of the synthetic identity fraud that rules based systems miss, and each alert carries a transparent risk score. False positive and precision figures are absent, which weighs more when alerts reach institutions that never saw the original evidence.

Grasshopper Bank's chief executive Michael Butler, BHG Financial's Andrew Stone and Digital Federal Credit Union's Gregg Stephens are quoted, and a FinCEN Tech Sprint win in the effectiveness and impact category is cited. No figure is joined to any of them.

Shared identities are encrypted under a key split across members, so no single institution can decrypt them, and only an institution already holding the matching record can validate a shared profile. Contributing intelligence does not mean contributing customers, a boundary a buyer can test.

Pattern matches and threat intelligence cross the network rather than consumer personal information, a product level handling rule stated in specifics. Its site privacy policy gives no retention periods or subprocessors for network data.

A SOC II badge appears with no type, period or auditor, and no trust portal is offered. The cryptographic design is described in detail, but a described mechanism is not an assessed control set.

FinCEN registration as a 314(b) association puts FiVerity inside the Patriot Act safe harbor that lets institutions share information on suspected money laundering and fraud without breaching customer confidentiality. That is an admission by the competent authority, not a self description.

Thin history, inconsistent data and sparse corroboration flag synthetic identities, and they also describe recent arrivals, young adults without credit and people rebuilding after disruption. A flag raised at one institution reaches every participant, and testing across population groups is unpublished.

Analyst review before any action gives a wrong match a point where it can be caught. Someone wrongly flagged as a synthetic identity may be marked across many institutions at once without being told, with no stated route for a correction to travel back, and no guarantee or indemnity is published.

Model providers, hosting providers and subprocessors are not named on the homepage, resource pages or privacy policy. Detection quality scales with active contributors, a number the vendor does not publish.

Ready in minutes without API integration, the platform works with existing tools, which removes the usual barrier for community banks and credit unions. Core banking, fraud and case management systems are not named, and no public API documentation is linked.

Data processing in the United States is stated in the site privacy policy, with no hosting provider, region or private deployment option for the network layer. Where network patterns physically sit is left unstated.

Price and basis of charge sit behind a contact form. Network products raise a question the pages leave open, whether contributors of intelligence pay the same as consumers of it.

Banks, credit unions and fintechs use the network, with regulators and law enforcement both contributing and receiving, and joint investigations are open to the roughly 46,000 institutions registered under section 314(b). Scope is domestic and limited to fraud and financial crime.