Cocos AI confidential-computing — attestation freshness bypass in AMD SEV-SNP handshake

Cocos AI's confidential-computing framework for running AI workloads inside AMD SEV-SNP trusted execution environments does not enforce freshness checks on attestation evidence when a specific configuration field is left nil, undermining the core trust guarantee of the confidential-computing setup.

Confidential-computing AI deployments exist specifically to protect sensitive model weights and data from the infrastructure operator; a flaw that lets stale attestation evidence pass verification defeats the entire security premise of the TEE-based AI workload isolation.

The intra-handshake attested TLS (aTLS) AMD SEV-SNP verification path fails to enforce attestation freshness when the expected reportData value is nil, allowing an attacker to replay or reuse stale attestation evidence to bypass the trusted execution environment's integrity guarantees.

Cocos AI ≤ 0.8.2

Upgrade to Cocos AI 0.8.2 or later; ensure reportData is always explicitly set in attestation verification configuration.