Your data stays in France. So do your models
ANSSI-qualified SecNumCloud 3.2 hosting, French AI models trained on French, end-to-end encryption, GDPR, HDS and AI Act compliance. No data is used to train third-party models.
Sovereignty, compliance, transparency
Three non-negotiable, verifiable, audited commitments. No vague cloud talk: we tell you where your data is, who can access it, and what we do with it.
Sovereignty
The whole chain — hosting, models, operations — stays French and European. No transfer outside the EU, ever.
- OUTSCALE / SecNumCloud 3.2 (Saint-Cloud, FR)
- Sovereign open-source models
- No US/CN Big Tech dependency
- Cloud Act-proof by design
Compliance
GDPR, HDS, AI Act: not in theory — in the contract. Signable DPA, up-to-date processing record, AIPD available.
- GDPR: DPA, record, AIPD
- HDS: healthcare hosting agreement
- AI Act: limited-risk classification
- ISO 27001: certification targeted 2026
Transparency
You know what the AI saw, what it wrote, and why. Every suggestion is traceable back to its source.
- Every annotation timestamped and signed
- Access logs available (RBAC)
- Sourced, verifiable citations
- No retraining on your data
Your documents don't leave France
End-to-end on sovereign infrastructure. From upload to report delivery, everything stays on SecNumCloud, encrypted, segregated, auditable.
Secure upload
HTTPS / TLS 1.3 · SSO or MFA auth · Encryption before send · No third-party cookies
AI processing
OUTSCALE SecNumCloud 3.2 · On-premises open-source models · AES-256 · Per-tenant isolation
Delivery
Digitally signed report · Complete audit trail · Deletion at D+30 · PDF + DOCX export
The detail that reassures your DPO
All contractual elements are available before signing. On request, our teams can help fill out your internal AIPD.
AI Act: ready before 2026
Eryon is classified "limited risk" under Regulation (EU) 2024/1689. Our systems embed transparency, traceability and governance obligations by design.
- Marking of AI-generated or modified content (art. 50)
- Technical documentation kept up to date (Annex IV)
- Guaranteed human oversight: AI proposes, humans validate
- Risk assessment refreshed at each model evolution
- Internal register of use cases and data used
Request our full security brief
12-page document: architecture, sub-processors, security measures, data governance, business continuity. To pass to your IT or DPO.