Security & compliance
The verification gate as a control surface
The verification gate is the platform's primary control for AI risk: AI output becomes authoritative only after a named human confirms it, and that decision is auditable.
What the gate guarantees
- An AI-derived obligation counts as authoritative only after a human verifier confirms it.
- An AI-derived standard position is promoted to a relied-upon position only after a human verifier confirms it.
- Every confirm, override and reject records the verifier identity, a timestamp, the before and after values, and a reason for overrides.
- Citations are preserved across overrides and rejections, so the source span is always recoverable.
What auditors can see
- The full extracted-versus-verified comparison for any obligation.
- Who verified what and when, scoped to the agreement, the user, or the period.
- Which obligations remain AI-proposed and await human confirmation, distinct from those already ratified.
- A record of the questions run through Ask clmAI.
Designed for assurance
If your risk function asks “how do we know the AI isn’t making things up?”, the answer is that AI output cannot become the relied-upon answer without a named human confirming it, and that confirmation is recorded in the verification history.