Emission gate
Gate
Stops unsupported claims before they reach the user. The Gate enforces release-or-block on every governed AI response, with the decision recorded in the Claim…
The Boundary Gate is the final emission checkpoint between the AI's generated response and any user, system, or downstream workflow. The Gate evaluates each claim against the verification record from the Claim Ledger, applies the deployment's gating policy (release, partial release, refusal, or escalation), and records the gating decision as a signed entry. Without the Gate, the Claim Ledger scores claims but nothing enforces the boundary — a response with one unsupported claim still reaches the user whole.
Without this: the Ledger scores claims but nothing enforces the boundary. A response with one unsupported claim still reaches the user whole. The scoring becomes telemetry, not governance.
How Gate turns a Ledger record into a release decision
Read this left to right from the Ledger handoff. A verified evaluation record enters, Gate applies deterministic policy, assigns an explicit response state, and either emits or blocks the result. Gate does not decompose claims, retrieve evidence, or generate new text.
How the Gate works
Reads the Ledger's evaluation record and makes a single deterministic decision about the entire response. Does not re-evaluate claims — enforces the boundary the Ledger defined.
- Reads per-claim scores, attribution, and gate reason codes from the Ledger
- Applies the output-state taxonomy based on aggregate claim support
- Assigns one of five states before the response reaches the caller
- Deterministic — same Ledger output always produces the same Gate decision
Who this is for
Governance runtime
executes automatically as the final step in the bounded-synthesis pipeline. No manual review, no override path.
The caller
receives a response already classified by the Gate — or an explicit state explaining why it was classified as partial, narrative, or blocked.
Gate — the emission policy boundary — is the deterministic emission checkpoint. Governance decisions made outside the model runtime are enforced here before claims are emitted to a user. Ledger — the integrity-protected inference audit trail — feeds the evaluation record. Gate is designed to stop unsupported decision-grade claims from being emitted as authorized.