Does Didit train on your data?

Their words · Privacy Policy

“ Didit trains, evaluates, and improves its identity-verification, biometric, and fraud-detection models, and operates cross-customer fraud-prevention safeguards, using anonymized or pseudonymized data derived from verification activity (for example: document features, fraud signals, attack patterns, model-error samples). Didit applies anonymization, pseudonymization, aggregation, and access controls so that the data used for training and fraud detection cannot reasonably be linked back to an identifiable individual outside the underlying verification record. This processing is grounded in Didit's legitimate interest in: improving the accuracy and safety of identity and fraud infrastructure used by all customers; detecting and preventing fraud, identity-theft attacks, deepfakes, and known-attacker repeat attempts; and meeting regulatory expectations around model performance, fairness, and security. Opt-out. A customer or end user may opt their data out of model training and fraud-detection processing by: deleting the underlying verification record via the API or the Business Console (the deletion removes the record from training pipelines on the next refresh cycle), or emailing privacy@didit.me with the relevant session identifier or account, requesting an opt-out. Opt-outs apply prospectively from the date of the request; Didit will also use commercially reasonable efforts to purge eligible records from active training datasets.”

Captured 2026-07-20. Open archived snapshot →Snapshot SHA-256: c76647e4473d37e9d292a8df885eb195d38b83057b5950d7acde9825f2e04663