Prompt
How do I store identity verification evidence for audits?
Latest observation
Nov 19, 2026
OpenAI APIWeb search: off
For audit-ready identity verification storage, the key is to keep what you need, for as long as required, with strong security and traceability, while minimizing sensitive data.
What to store
Keep an evidence package that can prove who was verified, when, how, and by whom/what:
- Customer identifier: internal user ID, application ID, case ID
- Verification result: pass/fail, risk score, decision, reason codes
- Verification method: document check, biometric match, database check, video KYC, etc.
- Evidence metadata:
- timestamp(s)
- verifier identity (agent/system)
- vendor/provider used
- policy/rule version applied
- Document references: document type, issuing country, last 4 digits only if needed
- Images/files only if required: ID scans, selfies, liveness video, OCR output, hash/signature
- Audit trail: every change, review, override, and re-verification event
What not to store unless required
Avoid storing unnecessary sensitive data:
- full ID numbers if not needed
- full DOB unless required
- raw biometrics unless legally justified
- excessive document images or duplicated copies
- notes containing sensitive personal data
Use data minimization and store only what your regulator, policy, or dispute process requires.
Security controls
Store evidence with strong protections:
- Encryption at rest and in transit
- Strict role-based access control
- Separation of duties for review vs. admin access
- Immutable audit logs or append-only event records
- Hashing/signing of files to detect tampering
- Backups and recovery testing
- Retention controls so records are deleted when no longer needed
Retention and deletion
Set retention based on:
- local AML/KYC regulations
- industry rules
- tax/accounting requirements
- legal hold obligations
- internal risk policy
Then define:
- retention period by evidence type
- deletion method
- legal hold procedure
- review schedule for stale records
Recommended storage model
A good pattern is:
- Operational record
Stores the verification outcome and metadata. - Evidence repository
Stores encrypted files or pointers to secure object storage. - Audit log
Records every access, update, and decision event. - Retention service
Automates expiry, archival, and deletion.
Example audit record
{
"case_id": "KYC-2026-00482",
"user_id": "u_918273",
"verification_type": "document_plus_selfie",
"result": "approved",
"provider": "VendorX",
"checked_at": "2026-11-19T10:15:00Z",
"reviewed_by": "agent_44",
"policy_version": "kyc-v3.2",
"evidence_refs": [
{
"type": "passport_image",
"storage_key": "evidence/2026/11/KYC-2026-00482/passport.enc",
"sha256": "..."
},
{
"type": "selfie",
"storage_key": "evidence/2026/11/KYC-2026-00482/selfie.enc",
"sha256": "..."
}
]
}
Practical best practices
- Store evidence in a separate, access-restricted system
- Prefer object storage with versioning and immutability
- Keep PII out of logs
- Use short-lived signed URLs for access
- Document your evidence retention policy
- Make sure you can produce an audit packet quickly for regulators
Important
If this is for a regulated process, you should confirm the exact retention and evidence requirements with:
- your compliance/legal team
- the applicable regulator
- your identity verification vendor
If you want, I can also give you:
- a database schema,
- a retention policy template, or
- a secure architecture diagram for storing KYC evidence.
Brands