Skip to Content
Security & TrustOverview

Security & Trust

VeriProof is built for regulated industries. Our security architecture combines cryptographic immutability, tenant isolation, and defense-in-depth controls so you can run high-stakes AI with confidence and retain evidence that third parties can verify.

🔐

Security contact: Discovered a vulnerability? Email security@veriproof.app. See our Responsible Disclosure policy.

What Makes VeriProof’s Security Different

Most AI governance platforms store audit logs in a traditional database. VeriProof adds multiple layers that competitors don’t:

LayerVeriProofTypical competitor
Audit log storageBlockchain-anchored (Solana CMT)Database only
Enterprise deployment boundaryCustomer-owned Azure subscription (Enterprise Federated)Vendor-managed shared environment
Multi-tenant isolationEF Core + PostgreSQL RLSApplication-layer only
Evidence verificationIndependent (no VeriProof dependency)Vendor-dependent
EU AI Act evidenceAuto-generated (Articles 9–17)Manual / partial
GDPR erasureCryptographic (blockchain-compatible)Delete row

Security Architecture

Blockchain Anchoring

Every AI session captured through the Ingest API is anchored to a Solana Concurrent Merkle Tree (CMT). This creates a tamper-evident record that:

  • Survives even if VeriProof’s infrastructure is unavailable
  • Can be independently verified by anyone with the proof hash

Blockchain Anchoring Deep Dive

Enterprise Federated data boundary

Enterprise Federated deployments keep session content, operational storage, and key custody inside your Azure subscription. VeriProof’s hosted infrastructure receives only the commitment data required for blockchain anchoring and verification workflows.

Enterprise Architecture

Multi-Tenant Isolation

Customer data is isolated using two independent mechanisms applied in tandem:

  1. EF Core Global Query Filters — every database query is automatically scoped to the requesting customer’s CustomerId at the ORM layer
  2. PostgreSQL Row-Level Security — a separate database-enforced policy that rejects cross-tenant access even if the application layer is compromised

Multi-Tenant Isolation Architecture

API Authentication

API keys use a compound format (vp_cust_{slug}.{azure-component}.{secondary}) backed by Azure Key Vault. Sandbox and production keys are distinct and cannot cross-contaminate.

API Authentication

Compliance Status

FrameworkStatusDetails
EU AI Act (Articles 9–17)✅ SupportedAuto-generated evidence packages
NIST AI RMF✅ SupportedSDK annotations map to all four functions
HIPAA / PHI✅ SupportedPII masking + audit trail controls
GDPR✅ SupportedCryptographic erasure, DPA available
SOC 2 Type II🔄 In ProgressTarget: Q4 2025
ISO 27001📋 Roadmap2026

Compliance Details  ·  Security FAQ

Infrastructure Security

  • Encryption at rest: AES-256 via Azure Storage service encryption
  • Encryption in transit: TLS 1.3 enforced on all endpoints
  • Key Management: Azure Key Vault for all signing keys and secrets
  • Secrets: Never hardcoded; bound to deployment context (fail-secure if context is missing)
  • Access Control: Explicit deny-by-default; no implicit wildcard grants

Encryption Details

Enterprise Security Features

FeatureSaaSEnterprise Federated
Data residency choiceAzure region selectionCustomer-owned Azure tenant
Network isolationVNet peeringFull private endpoints, no egress
Key custodyVeriProof-managedCustomer-owned Key Vault
Audit log accessVia portalDirect PostgreSQL + blockchain
Processing locationVeriProof-managed AzureCustomer-owned Azure subscription

Enterprise Architecture

Frequently Asked Questions

Can I verify a compliance record without going through VeriProof?

Yes. Every record includes a Merkle proof that can be verified against the Solana blockchain directly using standard Solana RPC tools. The verification algorithm is published on GitHub  so there is zero vendor dependency for audit evidence verification.

What happens to compliance records if VeriProof is unavailable?

Records anchored to the blockchain remain permanently accessible. The Solana CMT is a public, decentralized ledger. Your auditor can verify records independently of VeriProof’s infrastructure status.

Is our AI data ever used for training VeriProof models?

No. Customer data is never used for any purpose outside of providing the contracted service. See our Privacy Policy  and DPA.

How do you handle security vulnerabilities?

We operate a responsible disclosure program. Security researchers can report issues to security@veriproof.app. We aim to acknowledge reports within 24 hours and publish fixes within 90 days. See our full policy.

Last updated on