INNOVIUS.AI
Semantic encryption for the AI age: open-weight models that shield who you are
before any AI sees your text, with keys that stay yours.
Developed with the EECC Research Labs, trained at the Jülich Supercomputing
Centre — released for everyone.
You use the AI. The AI does not use you.
Compact encoder models (ModernBERT family) — the detection layer of ShinrAI semantic encryption. Privacy protection that works in multiple dimensions, across languages and document styles.
No black bars. Replacements keep culture, rarity and gender intact — so not even an AI can tell.
Classical tools redact ([REDACTED] wreckage) or hard-cut. ShinrAI substitutes are matched to context, culture and language at once — the AI reasons on, identities stay hidden.
Every substitute is statistically weighted so it never accidentally triggers biases or red flags downstream.
The matching key pair never leaves your device, your backend, or your instance. Only the key holder can map protected text back to reality.
The ShinrAI software optionally onion-routes requests, separating who asks from what is asked — across OpenRouter and any other access path. More paths, stronger protection.
De-identify electronic patient files (ePA) for pharma research, clinical archives, high-security environments and banking — data stays analytically useful.
Apache 2.0 weights; license-clean training-data releases (the fully synthetic corpus tracks license provenance per record — no proprietary APIs anywhere).
Need this protection now? A compact DE/EN/JA edition already runs in production — ChtSafe for individuals, the Secure AI Suite for organizations (on-premise first, SaaS-hosted on request).
Talk to usA compact edition of these models already protects production traffic in our commercial products — ChtSafe for individuals, the Secure AI Suite for organizations (on-premise first, SaaS on request).
Languages land as they pass our quality gates — the form below directly shapes the order. Tell us what you need.
We slot a limited beta cohort by language, hardware and domain, so every configuration gets real coverage.
Everyday AI, without giving yourself away.
Model checkpoints and quantized builds before the public release, matched to the hardware you tell us about.
Your language wishes, domains and votes feed directly into training priorities. This is genuinely how we decide what to build next.
Found a miss, a false positive, an awkward replacement? Beta feedback goes straight to the team training the next generation.
Everything lands as open weights right here on Hugging Face — beta participants simply get there earlier and shape what "there" looks like.
Two minutes helps us slot you into the best-fitting cohort. Or leave just an email and we will ping you once at release.
We will review your application and get back to you as we slot the next beta cohort. Meanwhile: follow Innovius on Hugging Face and @InnoviusAI — or talk to us if you need ShinrAI protection in production today.
This programme exists because remarkable institutions and open projects make serious research possible outside big tech.
The ShinrAI encryption models are developed by Innovius together with the research lab of the European EPC Competence Center (EECC) — long-time partners in applied AI and privacy research.
Training runs on the JURECA supercomputer, with the programme scaling onto JUPITER — Europe's first exascale system — at the Jülich Supercomputing Centre, supported through the WestAI initiative. Our deepest thanks to FZ Jülich and the JSC team: this work is only possible because Europe's research infrastructure is open to projects like ours.
The base encoder is mmBERT (Johns Hopkins CLSP, MIT) from the ModernBERT family. Training data is generated, cross-checked and evaluated by openly released models from the Qwen (Alibaba), Gemma (Google DeepMind), Mistral and NVIDIA Nemotron families — thank you for keeping frontier-quality open models available; this project runs no proprietary APIs at all.
Ground truth is anchored in open data: GeoNames (CC-BY), Wikidata (CC0), the US SSA & Census name statistics (public domain), and national open-data sources including INSEE (France), INE (Spain), the Polish PESEL registry statistics, and Italian & German municipal open data. Open data is what makes honest, bias-aware ground truth possible.
Open weights are a conviction, not a marketing angle. We fully support the Open Weights and American AI Leadership open letter published in July 2026 by an NVIDIA-led coalition of 50+ organizations: models whose weights anyone can download, inspect and run on their own infrastructure are defensive assets — for security, for competition, for trust. The ShinrAI encryption models are our contribution from Europe: open, inspectable privacy infrastructure you can run yourself.