CROVIA · Omission Oracle
Real-time AI Trust Scanner - Shadow Score for any model
Verifiable, cryptographically anchored public evidence of AI training data presence or absence. One observable fact. No model audits. No legal claims. Only presence/absence.
██████╗██████╗ ██████╗ ██╗ ██╗██╗ █████╗
██╔════╝██╔══██╗██╔═══██╗██║ ██║██║██╔══██╗
██║ ██████╔╝██║ ██║██║ ██║██║███████║
██║ ██╔══██╗██║ ██║╚██╗ ██╔╝██║██╔══██║
╚██████╗██║ ██║╚██████╔╝ ╚████╔╝ ██║██║ ██║
╚═════╝╚═╝ ╚═╝ ╚═════╝ ╚═══╝ ╚═╝╚═╝ ╚═╝
Crovia is the first cryptographically anchored trust layer for AI training data.
We don't audit models. We don't make legal claims. We observe one fact:
Is publicly auditable evidence present — or absent?
Real-time trust scanner for any HuggingFace model.
Enter a model ID → Get instant Shadow Score → See what's missing.
┌─────────────────────────────────────────────────┐
│ SHADOW SCORE │
│ │
│ 100 = All declarations present 🟢 GOLD │
│ 75 = Minor gaps 🟡 SILVER │
│ 60 = Needs attention 🟠 BRONZE │
│ <60 = Critical gaps 🔴 UNVERIFIED│
└─────────────────────────────────────────────────┘
277+ models tracked. Auto-updated every 6 hours.
We detect:
Crypto Evidence Packages — portable, offline-verifiable provenance for datasets.
Each capsule proves:
Add a Crovia badge to your model's README:

| Resource | Description |
|---|---|
| 🔮 Omission Oracle | Real-time model trust scanner |
| 🧾 CEP Terminal | Evidence verification console |
| 📊 Rankings | Global omissions leaderboard |
| 📦 CEP Capsules | Crypto evidence packages |
| ⚙️ Core Engine | Open-source engine |
| 🔐 PRO Engine | Enterprise features |
We don't trust promises.
We don't trust claims.
We trust evidence.
If you trained on data, prove it.
If you can't prove it, we see it.
There is nowhere to hide.