Skip to main content

privatemode.ai Integration

Overview​

privatemode.ai is a confidential LLM inference service that Enclava uses for processing sensitive data. It provides strong privacy guarantees through hardware-based trusted execution environments.

Key Features​

Privacy Guarantees​

  • No Data Retention: Conversations are never stored or logged
  • No Training on User Data: Your data is never used to improve models
  • Encrypted Processing: All inference happens within TEEs
  • Remote Attestation: Cryptographic proof of secure environment

Technical Implementation​

  • Runs on Intel SGX and AMD SEV trusted execution environments
  • End-to-end encryption from client to TEE
  • Stateless inference with no persistent storage
  • Regular security audits and compliance certifications

Security Flow​

  1. Request Encryption: Enclava encrypts user prompts before transmission
  2. Attestation Check: Verify privatemode.ai TEE status
  3. Secure Inference: Model processes data within TEE
  4. Encrypted Response: Results returned without data exposure

Compliance & Certifications​

  • GDPR compliant data processing
  • HIPAA eligible for healthcare workloads
  • SOC 2 Type II certification (in progress)
  • ISO 27001 compliance

Performance Characteristics​

  • Latency: 50-200ms overhead for TEE operations
  • Throughput: 100-500 tokens/second depending on model
  • Availability: 99.9% SLA for enterprise customers

External Resources​

For threat analysis, see privatemode.ai Threat Profile