Audit Detail

This is the hash-chained record. Each field is hashed and linked to the previous entry — tamper with one, the chain breaks.

Ledger entry

verity_audit_id

06ae6f91-b911-40e1-bd3d-7303ef78c57bVerify this entry →

model

gpt-4 (openai)

risk_tier

high

entry_hash

e4d804e97a2f94a113e1ae2eb1f018d28afe1b6427f450798db077f9bb27e1ae

created_at

2/27/2026, 2:18:05 AM

We just recomputed the hash from the raw data. If it matches, the record is untouched.

Integrity check

Chain valid: Yes

Recomputed entry_hash matches stored: Yes

This is the full reasoning trace: what was requested, which model was selected, what it returned, and whether validation passed.

Execution graph

{
  "graphId": "3b91ee9b-c20e-4c90-b1bc-26bbee0a599a",
  "release": {
    "released": true,
    "releasedAt": "2026-02-27T02:18:05.015Z"
  },
  "request": {
    "inputHash": "af3150898191f6d02f3aaace59fbf184e9cf4a6a912c719e85dfd95c6e9fa7c2",
    "receivedAt": "2026-02-27T02:18:05.015Z",
    "riskClassification": "high"
  },
  "validation": {
    "rulesApplied": [
      "policy-1"
    ],
    "failuresDetected": [],
    "fallbackTriggered": false
  },
  "modelExecution": {
    "latencyMs": 1805,
    "promptHash": "af3150898191f6d02f3aaace59fbf184e9cf4a6a912c719e85dfd95c6e9fa7c2",
    "rawOutputHash": "fd79920e2d026c5df73e9976e421beb193d8394762a36ab2193e78f5424bb6ba"
  },
  "modelSelection": {
    "reason": "risk-tier match",
    "policyVersion": "v1",
    "selectedModel": "gpt-4"
  }
}