Source · Tier B · DocumentationSource · Private Processing for WhatsApp: Technical White Paper and Security Guide
Private Processing for WhatsApp: Technical White Paper and Security Guide
Meta. 2026. Meta.
| Original | https://ai.meta.com/static-resource/private-processing-technical-whitepaper |
|---|---|
| Version | Version 2, updated 2026-03-16; version 1 was published 2025-06-10. |
| Accessed | 2026-10-15 |
| Authored or published by | Meta |
| Note | Meta's description of its own system; supports only "Meta reports" statements. Private Processing runs WhatsApp's AI requests in confidential VMs on AMD EPYC CPUs with SEV-SNP and NVIDIA Hopper GPUs. The client checks the attestation and transparency proofs, including entries in a third-party transparency log run with Cloudflare, before it connects. The threat model lists physical host attacks, with data-centre security, memory encryption and third-party relays as mitigations. |
Cited by
- TEE remote attestation for AI workloadsMechanism
- WhatsApp Private ProcessingImplementation
- MetaOrganization