Financial data scrubber
Pasting a bank statement, support ticket, or transaction log into an AI assistant can quietly expose account numbers, IBANs, and card details. This scrubber finds those financial identifiers and replaces them with masked placeholders — locally in your browser — so you can keep the useful context and drop the sensitive numbers.
What it detects
The scrubber targets the financial identifiers that most commonly appear in statements, emails, and support logs:
| Identifier | Format | Example masked as |
|---|---|---|
| IBAN | Up to 34 chars, country code + check digits | [IBAN] |
| Payment card | 13–19 digits (Luhn-validated in context) | [CARD] |
| BIC/SWIFT | 8 or 11 chars with institution + country code | [BIC] |
| UK sort code | XX-XX-XX or XXXXXX | [SORT-CODE] |
| US routing number | 9-digit ABA RTN in labelled context | [ROUTING] |
| Account number | Digits near the words “account” or “acct” | [ACCOUNT] |
A count per type tells you how many replacements were made.
How it works
Paste your text and the tool runs pattern matching for each identifier type. Account-number detection is deliberately conservative — it requires the word “account” or a similar label nearby — because bare runs of digits appear constantly in innocent contexts like order IDs, timestamps, and product codes. Tighter context means fewer false positives, but it also means an unlabelled account number can slip through. Read the scrubbed output before pasting it anywhere. Nothing is uploaded; all matching happens in your browser.
When to use it
The most common use case is preparing financial text for an AI assistant. You might want to ask about a transaction dispute, reconcile a statement, or troubleshoot a payment issue, but the raw text includes identifiers you should not hand to a third-party service. Paste the statement into this tool first, confirm the sensitive numbers are masked, then paste the clean version.
Other situations: redacting a screenshot description, sanitising a support ticket before logging it, or preparing an email for forwarding.
Tips for a clean result
- Verify before trusting. Scan the output for any digits that survived. An unlabelled account number or an unusual IBAN format can bypass the patterns.
- Context matters. A masked number combined with a name, sort code, and address can still identify someone — when in doubt, delete surrounding context too.
- Pair with the SSN / national ID scrubber. Bank letters and statements often include national identity numbers alongside financial identifiers.
- Copy the scrubbed version only. Never paste the original — scrub first, paste second, every time.