P233 — AIEP — Evidence-Bound Simulation Certification Engine
Applicant: Neil Grassby Classification: Patent Application — Confidential Priority: Claims priority from GB2519711.2 filed 20 November 2025 Architecture Layer: AIEP Phase 2 Support Layer
Framework Context
[0001] This specification operates within an AIEP environment as defined in GB2519711.2 and GB2519798.9. The present specification defines a simulation certification mechanism that issues verifiable proofs of simulation validity for counterfactual reasoning outputs, enabling downstream consumers to verify that simulation conclusions are grounded in admitted evidence.
Field of the Invention
[0002] The present invention relates to simulation certification and verifiable computation proofs for counterfactual reasoning in evidence-bound artificial intelligence.
Background
[0003] Counterfactual reasoning engines (P203, P204) produce simulation outputs that represent hypothetical world states. These outputs are used by goal arbitration, intervention planning, and strategy systems. Without a certification mechanism, consumers of simulation outputs cannot distinguish evidence-grounded simulations from speculative or corrupted outputs.
Summary of the Invention
[0004] The invention provides an Evidence-Bound Simulation Certification Engine (EBSCE) that: receives the evidence activation set used by a simulation run; constructs a hash-linked certificate recording the root world state, the activated evidence, the simulation parameters, and the hypothesis applied; signs the certificate with the system identity key; and appends the certificate to the simulation output.
[0005] Simulation consumers may verify a certificate by replaying the simulation from the certified inputs and confirming output equivalence. The certificate chain is recorded in the evidence ledger, enabling ledger-level auditability of simulation claims.
ASCII Architecture
Simulation Run Complete (P203/P204)
|
v
+--------------------------------------------+
| Evidence-Bound Simulation Certification |
| Engine (EBSCE) |
| |
| Collect evidence activation set |
| Record root world state (CWSG hash) |
| Hash simulation parameters |
| Construct certificate record |
| Sign certificate |
| Append to evidence ledger |
+--------------------+------------------------+
|
v
Certified Simulation Output
Certificate Hash → Ledger Admission
Detailed Description
[0006] Certificate Construction. For each simulation run, the EBSCE constructs a certificate containing: the SHA-256 of the CWSG root state at simulation start; a Merkle root of all evidence artefact hashes in the activation set; the simulation parameters (hypothesis statement, branch factor, depth limit); a hash of the simulation output; and a timestamp.
[0007] Signature. The certificate is signed using the system’s private identity key. This ensures the certificate cannot be forged or transferred between systems without detection.
[0008] Ledger Admission. The signed certificate is admitted to the evidence ledger as a first-class artefact, giving it the same append-only, hash-linked integrity guarantees as evidence artefacts.
[0009] Replay Verification. Any authorised consumer with access to the original evidence artefacts may replay the simulation from the certified inputs. If the replay output hash matches the certified output hash, the simulation conclusion is verified.
Technical Effect
[0010] The invention provides cryptographically verifiable simulation certification for counterfactual reasoning outputs in evidence-bound AI systems. By recording the exact evidence activation set and root world state hash in a signed certificate admitted to the AIEP evidence ledger, the engine enables any authorised party to independently verify a simulation conclusion by replay, without access to the original running system. By making simulation certificates first-class evidence artefacts, the engine integrates simulation auditability into the broader AIEP evidence integrity framework.
Claims
-
A computer-implemented method for evidence-bound simulation certification, the method comprising: (a) capturing the input set of a counterfactual simulation comprising: the evidence artefact activation set, the root world state hash at simulation start, and the simulation parameter configuration; (b) executing the simulation and capturing the output hash of the simulation conclusion; (c) constructing a simulation certificate comprising: the input set hash, output hash, simulation execution timestamp, and system identity; (d) signing the certificate with the system’s private identity key; and (e) admitting the signed certificate to the AIEP evidence ledger as a first-class evidence artefact.
-
The method of claim 1, wherein certificate validity is verifiable by replay: retrieving the certified evidence activation set, re-executing the simulation from the certified root world state, and confirming the output hash matches the certified output hash.
-
The method of claim 1, wherein the certificate is verified by the Counterfactual Timeline Engine before a simulation conclusion may be used to support claims in a reasoning session.
-
The method of claim 1, wherein post-issuance modifications to any source evidence artefact in the activation set invalidate the certificate, detected by recomputing the activation set hash.
-
The method of claim 1, wherein simulation certificates are linked to the associated reasoning session evidence artefact by cross-reference identifier, enabling complete session audit trail traversal.
-
An Evidence-Bound Simulation Certification Engine comprising: one or more processors; memory storing a certificate store, signing key, and verification replay interface; wherein the processors are configured to execute the method of claim 1.
-
A non-transitory computer-readable medium storing instructions that, when executed by a processor, implement the method of claim 1.
Abstract
An evidence-bound simulation certification engine for counterfactual reasoning systems captures the exact evidence activation set and root world state at simulation start, hashes the simulation output, constructs a signed simulation certificate, and admits the certificate to the AIEP evidence ledger as a first-class artefact. Simulation conclusions can be independently verified by any authorised party through deterministic replay. Post-issuance modification of source evidence invalidates the certificate through activation set hash mismatch.