Remote ID verification tech is often biased, bungling, and no good on its own

A pre-press US GSA study found that five remote identity verification (RiDV) systems are unreliable and inequitable across demographic groups, with one product failing around 50% of the time and the best still failing 10%; vendors were anonymized, though a prior privacy document lists TransUnion, Socure, Jumio, LexisNexis, Incode, and Red Violet. While login.gov is using a high-performing algorithm per NIST FRVT evaluations, the final peer-reviewed study is due in 2025 with deeper analysis of error causes, and LexisNexis advocates a multi-layered approach beyond facial recognition and IAL2 to reduce risk and improve equity.