Running AI training or inference inside TEEs, so prompts, data and model weights stay protected in use, and clients can verify the service by attestation.
Confidential AI combines CPU TEEs with GPU confidential computing. The client, or a verifier it trusts, checks attestation evidence for both before sending data or releasing keys.
Models, drivers and serving code change often. Each change alters what is measured, so confidential AI depends on re-attestation and on reference values that keep pace with releases.
Example. A law firm uses a hosted language model. Its documents are decrypted only inside attested hardware, and the firm's own tool checks the evidence first.
Working on re-attestation for your product? A first call takes 30 minutes and commits you to nothing. Bring your product and the one question that worries you.
Next fixed date: EU Digital Identity Wallets are due in every member state on 24 December 2026, 42 days from now.