For Compliance Officers ·
What you'll accomplish
You'll build a systematic ChatGPT workflow for drafting Suspicious Activity Report (SAR) narratives that cuts the writing time from 1–3 hours to 15–20 minutes of AI-assisted drafting plus review. The workflow uses sanitized activity descriptions (no real customer data) to produce structurally correct, FinCEN-compliant narratives.
Privacy rule — non-negotiable: Never include real customer names, account numbers, Social Security numbers, dates of birth, or any other personally identifiable information in any AI prompt. Use generic descriptions only. Add real details only in your internal SAR filing system.
What you'll need
Every SAR narrative must answer these questions:
Open ChatGPT and start a new chat. Paste this opening context:
I'm a bank compliance officer drafting SAR (Suspicious Activity Report) narratives. I will describe suspicious activity using only generic terms — no real customer names, account numbers, or personal information. Please draft SAR narratives using FinCEN's required 5-part format: (1) who the subject is, (2) what activity occurred, (3) when/how often, (4) where (transaction channels), (5) why it's suspicious. Use formal, factual, third-person narrative language. Keep each narrative between 150-300 words.
For each SAR, describe the activity using only generic terms:
Example of CORRECT sanitized description:
Activity to describe: A sole proprietor business account received 12 structured cash deposits over 30 days, each between $9,000 and $9,900. The business is a cash-intensive retail operation. No single deposit exceeded the $10,000 CTR threshold. The deposit pattern is inconsistent with the business's typical deposit history. No apparent business reason for structuring.
Example of INCORRECT (never do this):
John Smith's account #123456789 at our Main Street branch received...
Before finalizing: "Does this narrative clearly explain why the activity is suspicious, or does it just describe the activity without explaining the red flags?" Ask ChatGPT to strengthen any section where the "why suspicious" reasoning is weak.
For structuring:
SAR narrative for: cash deposits structured to avoid CTR reporting threshold. [Add your generic activity description]. Include: the pattern, the threshold reference, and why this is inconsistent with normal activity.
For unusual wire activity:
SAR narrative for: unusual international wire transfers inconsistent with account purpose. [Add generic description]. Include: origin/destination countries (generic), amount ranges, frequency, and the red flags.
For account takeover:
SAR narrative for: suspected account takeover fraud. [Add generic description of activity — no real data]. Include: what changed in the account behavior, how fraud was detected, and what actions were taken.