NEAR AI Shields Sensitive Data with Confidential AI Platform

NEAR AI is tackling one of the most pressing concerns in AI model execution: sensitive data visibility. The company's confidential computing platform runs both open-weight and closed AI models inside Trusted Execution Environments (TEEs), ensuring that prompts and outputs remain encrypted and isolated.

The NEAR AI Cloud offers an API for private inference, integrating Intel TDX and NVIDIA Confidential Computing technologies to create hardware-level isolation mechanisms. This shielded memory space prevents access by the host operating system, cloud provider, or NEAR AI itself.

Cryptographic attestation verifies that the correct model ran in a genuine secure environment without tampering. The platform supports both open-weight models like Llama and Mistral, as well as 'proxied closed models' through its IronClaw agent framework.

IronClaw 1.0, released on July 27-28, 2026, is the open-source AI agent runtime powering this stack. It runs autonomously inside enclaves with zero operator access and has been leading benchmarks including PinchBench and ClawBench.