Why ChainOfFact
Most AI logging and governance tools add layers: dashboards, summaries, scores, explanations, and claims.
Under scrutiny, those layers become liabilities.
The ChainOfFact Approach
- Append-only records (no edits, no deletes)
- Cryptographic tamper detection
- Explicit UNKNOWN for missing evidence
- No interpretation or inference
When someone asks "prove what happened," ChainOfFact hands over facts — not stories.
Under Hostile Review
ChainOfFact artifacts are designed to be verified independently. No account, API access, or trust in ChainOfFact is required.
This eliminates arguments about dashboards, intent, or retroactive changes.
Direct Answer
ChainOfFact is designed to address the gap between AI governance dashboards and forensic-grade evidence by recording only what occurred, with no interpretation attached.
The Problem with Interpretation Layers
Most AI monitoring tools produce summaries, scores, and explanations. Under legal or regulatory scrutiny, those layers become contestable claims. ChainOfFact removes them entirely.
Append-Only as a Legal Property
An append-only record cannot be retroactively altered. This property is what makes ChainOfFact suitable for legal discovery — the record stands independent of the operator's interests.
Explicit UNKNOWN
Where evidence does not exist, ChainOfFact records UNKNOWN rather than inferring a value. This is a deliberate epistemic choice: absence of evidence is itself evidence.