Anti-Money Laundering

Automated monitoring and investigation of transactions, customers, and alerts to detect money laundering risk and reduce false positives.

Business impact

• False positives — HSBC reported a 20% reduction, and Holvi reported cutting the time spent remediating false positives in half.
• Customer conversions — Pockit reported a 15% increase after automating identity verification.
• Processing time — The Nearform case reported 847 cases processed in under two hours, versus about a year of human effort.
• Manual investigation effort — The Nearform case said the batch would have required close to 30 additional staff to handle at the same volume.

Data requirements

• Transaction data (tabular) — Used for AML monitoring and anomaly detection in the HSBC, Holvi, and ClearBank cases.
• Customer onboarding data (tabular) — Used for identity and AML checks at account opening in the Pockit and Holvi cases.
• Government identity documents (image) — Used by Onfido to verify legal identity.
• Selfies (image) — Used by Onfido to match a person to their identity document.
• Sanctions and watchlist data (tabular) — Used in Holvi’s screening and monitoring process.
• Adverse media and risk databases (text) — ComplyAdvantage used structured and unstructured risk sources to identify financial crime risk.

AI methods and techniques

• Machine learning — Used to monitor transactions, identify anomalous activity, or automate identity checks.
• Unsupervised learning — Ayasdi’s platform used unsupervised methods to find unique groupings and relationships in transaction data.
• Decision tree — Ayasdi used decision trees to show how decisions are made and help avoid black-box outputs.
• Feature engineering — Used in the HSBC case to prepare data for model validation and interpretation.
• Behavioral analytics — Featurespace’s ARIC platform built individual behavioral profiles and detected abnormal activity in real time.
• Facial recognition — Onfido used it to match selfies with identity documents for remote verification.

AI models and model families

Ayasdi, ComplyAdvantage platform, Onfido, Featurespace ARIC

Industries

Real-world evidence

5 documented case studies on record.

Companies using this: financial services arm of a state-owned service provider, HSBC, Holvi Payment Services, Pockit, ClearBank.