@NEARProtocol has handed verification of its confidential AI workloads to @intel's Trust Authority, shifting the chain of trust away from the platform operator and anchoring it directly at th
@NEARProtocol has handed verification of its confidential AI workloads to @intel's Trust Authority, shifting the chain of trust away from the platform operator and anchoring it directly at the chipmaker level.
NEAR AI Cloud has run on Intel and NVIDIA confidential computing hardware since December 2025. What is new is who validates the proofs. Every workload now returns an Intel-signed attestation token confirming that the job ran inside sealed, tamper-resistant hardware. @NEARProtocol operates the service, but @intel independently checks it, meaning users no longer have to take the operator's word for it.
That distinction is significant. Intel Trust Authority decouples the roles: Intel itself is the appraiser, the host is only the host, and the relying party receives a token signed by an independent third party with no operational stake in the workload. This resolves a long-standing weakness in confidential computing deployments, where infrastructure providers have historically relied on self-attestation.
Intel Trust Authority is a zero-trust attestation service that helps verify apps, data, and workloads are running in trusted environments. It validates infrastructure, applications, and AI workloads using Intel confidential computing Trusted Execution Environments (TEEs) on CPUs and GPUs.
NEAR AI Cloud delivers private inference in three steps: a user's prompt is encrypted locally and transmitted to a Confidential Virtual Machine (CVM) inside a TEE; processing occurs inside a Trusted Execution Environment that prevents external access, including from NEAR itself; and inside the enclave, the model decrypts the input, performs inference, and re-encrypts the result before returning it.
By running models in TEEs with cryptographic attestation, the platform creates an auditable chain of trust. Organizations can verify not just that their data was encrypted in transit, but that it remained encrypted during processing and that the model executing the inference was the one they expected.
The integration is live now. For enterprises handling sensitive data, the move addresses a core concern: one misconfiguration or insider action can break traditional software-level trust signals instantly, which is why the AI industry needs verifiable privacy, not unsupported promises.
Sources:NEAR AI: Introducing NEAR AI Cloud and Private ChatIntel: Intel Trust Authority OverviewCrypto Briefing: NEAR AI Builds Infrastructure for Private AI Model Execution
