Governance Execution Certificate · Patent Pending · GB2607087.0
Proof that governance occurred before delivery
A per-inference cryptographic certificate proving that a specific AI output passed a non-bypassable governance evaluation before it reached any user, system, or regulator. Externally verifiable. Architecturally unforgeable. Issued only after aiGUARD issues an execution permission.
GEC · Live Certificate Specimen
gec_idgec_7f3a9c2d-e814-4b1a-9f2e-3c7d8a1b4e5f
execution_decisionALLOW
confidence_score94
confidence_bandGREEN
consequence_levelMEDIUM
user_state_hashsha256:a3f7c9e2b1d4...
output_hashsha256:8b4e2f1a9c3d...
accessibility_complianttrue
issuerhttps://gec.aiguard.systems
timestamp2026-04-14T09:31:07.412Z
How GEC Works
Four steps. Every inference. No exceptions.
Step 01
aiGUARD evaluates the output
aiGUARD evaluates every candidate AI output against confidence, consequence, and user state. A runtime execution permission is derived. Delivery remains blocked until that permission is issued.
Step 02
GEC is issued on permission
Only after aiGUARD issues an execution permission does GEC generate a certificate. The certificate cannot be created by any external caller. It cannot be generated outside the governance pipeline. It is architecturally unforgeable.
Step 03
Output and certificate are bound
The certificate is cryptographically bound to the delivered output via output hash. The governance parameters applied to that specific output are recorded in full, including confidence, consequence, user state, execution decision, and accessibility compliance.
Step 04
Any third party can verify
GEC certificates are externally verifiable via a public endpoint. Any regulator, auditor, or downstream system can verify that governance occurred, without accessing internal systems, without trusting the operator's word, without any platform access at all.
Architectural Guarantees
What makes GEC unforgeable
GEC is not a signed receipt that any system can produce. Its unforgeability is architectural, built into the conditions under which it can be generated.
- Certificate issued only after a completed execution permission for a specific output, it cannot precede governance
- Certificate records the actual governance parameters applied: confidence, consequence, user state hash, execution decision, output hash, accessibility
- Certificate generation uses an authenticated internal interface with hardware-protected signing, no external caller can invoke it
- Certificate cannot be generated outside the governance pipeline, architecturally impossible, not merely prohibited by policy
- Any third party can verify via public endpoint without platform access, the verifier does not need to trust the operator
- If certificate generation fails, delivery is blocked, no certificate means no delivery, always
The aiGUARD Trust Stack
Three products. Complete governance.