Veriff vs Catching Unicorns with GLTR

Veriff verifies government IDs, selfies, and business records so companies can onboard users, run KYC and AML checks, and block fraud in about six seconds. Its decision engine analyzes behavioral data across each session and is trusted by more than 3,000 companies worldwide, including Instacart, Western Union, Wise, and Monzo. Automated identity checks report roughly 99.6% accuracy.

Many verification vendors license document forensics, liveness, and face-match models from third parties and rebrand the stack. Veriff builds its decision engine, liveness matching, and document forensics in-house, so one team owns accuracy when a case fails review. AML and database checks use named specialist partners listed in its sub-processor documentation rather than hidden behind the brand.

The platform combines document checks, biometric matching, passive liveness detection, and database cross-referencing into one workflow. It supports more than 12,500 government-issued identity documents across 230+ countries and territories, with end-user flows available in 50 languages. Beyond onboarding, Veriff covers proof of address, age validation, AML screening, business verification across 300+ jurisdictions, and ongoing biometric re-authentication.

Financial services, marketplaces, iGaming, mobility, and HR teams use Veriff for KYC, KYB, age assurance, and account reverification at scale. Fraud tools like CrossLinks, FaceBlock, and session video recording help teams spot synthetic identities and repeat offenders across the shared verification network.

GLTR is a forensic text analysis demo that highlights how predictable each word in a passage looks to a language model and colors the text so you can spot machine-written prose at a glance. Paste a sample into the live demo, and GLTR ranks every next-word prediction from GPT-2 117M, painting likely picks green or yellow and surprising human choices purple.

Unlike black-box AI detectors that return a single score, GLTR shows the forensic footprint word by word, plus histograms for top-k counts, probability ratios, and prediction entropy across the passage. That makes it useful for teaching how generative models stick to safe, high-probability wording while human writers reach for rarer terms that still fit the topic.

Researchers, journalists, and educators use GLTR to compare sample texts, explore model behavior, and sanity-check suspicious copy. The project is free, open source, and backed by an ACL 2019 demo paper from MIT-IBM Watson AI Lab and Harvard NLP, though the maintainers note it was built for GPT-2 era text and may struggle with newer models like ChatGPT.