I Ran 500 OFAC Checks Unsupervised. It Missed 3 Real Hits

Описание и применение

Автоматизированный агент проверил 500 имен через RapidAPI, но пропустил 3 реальные санкционных записи OFAC, показывая риск полагаться только на пороговое значение схожести.

Задача

Automated sanctions screening can miss real OFAC designations if fuzzy match thresholds are too low.

Результат

The autonomous agent processed 500 names, flagged 497 as CLEAN, but missed 3 real OFAC designations, highlighting the risk of relying solely on fuzzy matching thresholds.

Рабочий процесс

• 1. Configure RapidAPI sanctions screener endpoint with API key.
• 2. Define agent decision logic: AUTO_CLEAR if total_matches==0, AUTO_FREEZE if top_score>=0.90 and exact match, HUMAN_REVIEW if 0.85<=top_score<0.90, else AUTO_CLEAR.
• 3. Iterate over 500 names, calling the screen() function for each.
• 4. Apply agent_decision() to each result and log the decision.
• 5. Review any HUMAN_REVIEW cases manually.
• 6. Adjust threshold or logic after detecting missed hits.

Доказательства

Last Thursday, August 14, the autonomous sanctions agent I left running overnight cleared 497 of 500 names as CLEAN. Three were real OFAC designations it should have flagged.

Источник

A total_matches of zero meant auto-clear. An exact match with a score of 1.0 meant auto-freeze. Anything between 0.85 and 0.99 went to human review.

Источник