GBG processes over 62 billion identity transactions annually, running machine learning models against global data aggregates to separate legitimate users from fraudsters at scale. The threat model is customer onboarding - where financial services, healthcare, and e-commerce platforms face synthetic identity attacks, account takeover attempts, and document fraud. GBG's platform powers 800 million identity checks across 90 countries, using biometric verification, address validation, and real-time risk scoring to block bad actors while clearing genuine customers in seconds. Clients include HSBC, Visa, ASOS, and IBM.
The technical stack centers on Python, C++, and deep learning frameworks - PyTorch, TensorFlow, Keras - for training fraud detection models that adapt to evolving attack patterns. OpenCV handles document and biometric analysis; JAX accelerates model inference at transaction volume. Backend infrastructure runs on Microsoft Azure with PostgreSQL databases, processing identity verification requests that touch data sources spanning fifteen countries of operation. The platform maintains a single authoritative record for identity and address data, enabling deterministic verification decisions under regulatory constraints like KYC and AML requirements.
Founded in 1989 and now a FTSE 250 constituent, GBG employs over 1,100 people and serves 20,000+ organizations. The operational focus is global data orchestration - aggregating identity signals from disparate sources, normalizing formats, and running inference pipelines that must handle both high-assurance scenarios (opening bank accounts) and lower-friction use cases (e-commerce checkout). The company's longevity reflects the sticky nature of identity infrastructure: once embedded in a bank's onboarding flow or a retailer's fraud stack, these systems become difficult to replace.