Compliance, with a receipt.
Most "AI compliance" stops at access logs and prompt-level guardrails. Divinci goes inside the model. Every entity association is a queryable feature; every edit is a portable, auditable patch — and the vIndex is the technical-documentation artifact your regulator is about to start asking for.
Watch a fact leave the weights
Ninety seconds: what a vIndex is, how a rank-1 patch removes one association from a frozen model, and what the receipt looks like.
The worked example in full: Deleting Paris from a Language Model
What the regulation requires
The same artifact answers transparency, oversight, documentation, and erasure requirements.
| Regulation | Requirement | Competitors | vIndex |
|---|---|---|---|
 EU AI Act Annex IV | Technical documentation auto-generated from the vIndex | Access logs + prompt guardrails only | Annex IV dossier generated from release records and the model-edit log — with every gap stated |
| GDPR Art. 17 | Right-to-erasure with verifiable proof | Soft-delete, no proof | Rank-1 patch erases the association; receipt issued |
| HIPAA §164.312 | Audit trails for protected health information | Manual audit logging | Every entity association is a queryable feature |
 NIST AI RMF MAP/MEASURE | Risk management framework mapping | Static controls checklist | Continuous, model-native evidence |
 SOC 2 Trust Services | Security, availability, confidentiality | Point-in-time audit | Deployment telemetry as evidence input — Divinci is not SOC 2 certified |
 ISO/IEC 42001 AI Management | AI management system standard | Document-only conformity | vIndex as the system of record — certification not yet held |
What a "receipt" actually looks like
A vIndex patch is a portable JSON file with a SHA-256 checksum. Apply it to suppress a fact; remove it to restore the model bit-for-bit. The patch IS the audit log — operators, regulators, and downstream consumers can verify the same operation independently.
Below: an illustration of the format — the fields a Gate-3 DELETE patch carries, and the checksum that makes it verifiable. It is an example of the shape of a receipt, not a measurement we are publishing here.
{
"name": "gdpr-art17-paris-capital",
"version": 1,
"base_model": "google/gemma-4-E2B-it",
"created_at": "2026-04-22T22:34:00Z",
"operations": [{
"op": "delete",
"entity": "Paris",
"relation": "capital",
"target": "서울",
"weight": -1.0,
"layer": 27,
"feature": 11179
}]
}Bring an erasure request. Leave with the dossier.
Pick a single fact you need removed from a model you run. We apply the edit, run the probe, and hand back two artifacts: the signed vIndex patch that performed the erasure, and an EU AI Act Annex IV dossier assembled from your release records — with every section we could not complete named explicitly.
Two limits worth stating up front. Weight-level erasure requires an open-weight served model — GPT, Claude and Gemini cannot be edited this way, only retrieved-context can. And the dossier is a draft for your assessment, not a declaration of conformity: that is a judgement only you can make.
Gemma
Qwen
Llama
Mistral
Microsoft
Twelve published vIndexes — Browse them on Hugging Face
Compliance you can prove. On any open transformer.