title: README
emoji: 🔐
colorFrom: indigo
colorTo: blue
sdk: static
pinned: false
Innovius — Secure AI, built for trust
Innovius builds the Secure AI Suite: private, governed AI for enterprises — on-premise, VPC or air-gapped — powering ChtSafe and AICNCT. At its heart sits ShinrAI (信頼 — Japanese for trust): a transparent, realtime semantic privacy layer for AI communication.
🧪 ShinrAI Encryption Models — Open Beta
*A joint development by Innovius × EECC Research Labs, trained at the Jülich Supercomputing Centre.*
You use the AI. The AI does not use you. We are publishing our next-generation privacy models here: open-weight semantic encryption models (ModernBERT family), compact enough to run on your own hardware. Classical PII tools redact or hard-cut; ShinrAI encrypts semantically — replacements matched to context, culture and language, statistically weighted so they never accidentally trigger biases or red flags. The AI reasons over natural text; the identities inside were never exposed. In the full ShinrAI stack the matching key pair never leaves your device, backend or instance, with optional onion routing across as many model access paths as you like (OpenRouter and friends — the more, the stronger). The same models protect data at scale: de-identifying electronic patient files (ePA) for pharmaceutical research, clinical archives, high-security environments and banking datasets.
| Status | Languages |
|---|---|
| 🟢 Available today | German, English, Japanese — a compact edition already protects production traffic in ChtSafe and Secure AI Suite deployments (contact us for the enterprise version: on-premise first, SaaS-hosted on request) |
| 🟠 In training | The full open-weight suite: Italian, French, Spanish, Polish, Portuguese (PT/BR), Russian, Ukrainian, Turkish, Korean, Arabic, Dutch, Swedish, Chinese (Simplified) — and your vote decides the order |
Apache 2.0 open weights. License-clean training-data releases. No proprietary APIs anywhere in the pipeline. The training corpus is synthetic — generated and quality-checked by openly released models, anchored in public registries and open data. Every record carries license provenance, so the data we release is clean by construction.
👉 Apply for beta access — or get notified at release
Beta participants get early checkpoints matched to their hardware, and their language, domain and quantization votes directly shape our training roadmap. You can also follow this organization to see every release the moment it lands.
Partners & thanks
Developed together with the EECC Research Labs (EECC on Hugging Face) — our closest research partner and co-developer of the ShinrAI encryption models, not a subcontractor.
Trained on the JURECA supercomputer at the Jülich Supercomputing Centre (Forschungszentrum Jülich), scaling onto JUPITER, Europe's first exascale system — supported by the WestAI initiative. Our deepest thanks to FZ Jülich and the JSC team: work like this is only possible because Europe's research infrastructure is open to it.
Training data is generated, cross-checked and evaluated entirely by openly released models — the Qwen (Alibaba), Gemma (Google DeepMind), Mistral and NVIDIA Nemotron families — on the base of mmBERT (Johns Hopkins CLSP) from the ModernBERT lineage, anchored in open data: GeoNames, Wikidata, and the statistics offices and open-data portals of the countries we cover. Thank you for keeping frontier-quality open models and open data available.
Open weights, on principle
We fully support the Open Weights and American AI Leadership open letter published in July 2026 by an NVIDIA-led coalition of 50+ organizations: open weights are defensive assets — for security, for competition, for trust. The ShinrAI encryption models are our contribution from Europe.
Links
🌐 innovius.ai · 🔐 ShinrAI technology · 🧪 Model beta · ✉️ Contact · 🐦 @InnoviusAI