97% F1 on adversarial Indian PII samples

AI PII Detection for Indian Data – Aadhaar, PAN, UPI & More

CrewCheck's multi‑layer detection uses Verhoeff checksums, Hinglish normalisation, and contextual AI to catch what regex misses.

What we detect

Every Indian PII type your AI might leak

12-digit UID with Verhoeff checksum validation. Detects spaced, hyphenated, and word-digit obfuscated variants.

10-character alphanumeric. Catches obfuscated formats like 'A B C D E 1 2 3 4 F' and Hinglish mixed text.

Virtual Payment Addresses (user@bank). Detects obfuscated 'dot at' variants common in Hinglish prompts.

+91 and 10-digit Indian mobile numbers. Context-aware to avoid false positives on order IDs.

14-digit Ayushman Bharat Health Account IDs. Validated with healthcare context signals.

IFSC codes and 9–18 digit bank account numbers with context-aware filtering.

How it works

Pattern matching with mathematical checksum validation (Verhoeff algorithm) for Aadhaar. Eliminates false positives from random 12-digit numbers.

Converts word-digit obfuscation ('ek do teen char…') to numeric form before detection. Catches what pure regex misses.

Semantic analysis checks surrounding context (e.g., 'kyc', 'aadhaar', 'payment') to confirm PII intent and suppress false positives.

Replacements applied from end to start of string so character indices never shift. Full prompt context preserved for downstream LLMs.

Paste any prompt containing Indian PII into the free scanner. No sign-up, no SDK, no configuration.

Example input

Customer Aadhaar: 2345 6789 0124
PAN: ABCDE1234F
UPI: harsh@okicici
Mobile: 9876543210

Redacted output

Customer Aadhaar: [AADHAAR]
PAN: [PAN]
UPI: [UPI]
Mobile: [MOBILE]