# Linux Foundation takes governance of TRACE, an open standard for AI runtime attestation > Source: > Published: 2026-08-26 05:24:29+00:00 Photo: igovar igovar / Pexels # Linux Foundation takes governance of TRACE, an open standard for AI runtime attestation The new specification produces cryptographically verifiable records proving AI workloads ran as promised, backed by silicon-level attestation from AMD and Intel hardware. If you’ve ever wondered how to prove that an AI agent actually did what it claimed to do, on the hardware it claimed to use, with the data policies it claimed to follow, you’re asking the right question. The Linux Foundation just announced it will govern an open standard designed to answer it. TRACE, which stands for Trust, Runtime Attestation and Compliance Evidence, is an open specification that generates hardware-attested, cryptographically verifiable records of AI agent and confidential workload execution. Think of it as a tamper-proof receipt for every AI operation, one that a third party can verify without having to take the operator’s word for anything. ## What TRACE actually does The specification was originally developed by OPAQUE, a confidential computing firm, in collaboration with AMD, Intel, Microsoft, and the Technology Innovation Institute. Rather than building an entirely new framework from scratch, TRACE integrates several established standards including RATS (RFC 9334), EAT (RFC 9711), SLSA, SCITT, SPIFFE, and EAR to ensure interoperability across different cloud platforms and secure computing environments. The core output is what the project calls “Trust Records.” These documents capture operational details like the runtime environment, the software that was executed, data classifications, and the policies that were enforced. All of it is secured by silicon attestation mechanisms, meaning the verification is rooted in hardware rather than software promises. In practical terms, TRACE leverages hardware attestation technologies like AMD SEV and Intel TDX. TRACE essentially gives those capabilities a standardized language to speak when it comes to AI workloads. The specification was first presented at the Confidential Computing Summit on June 23, 2026. In the roughly ten weeks since, its reference implementation has been downloaded approximately 135,000 times on PyPI, the Python package repository. ## Why neutral governance matters here Jim Zemlin, CEO of the Linux Foundation, emphasized the importance of neutral governance for creating verifiable AI trust records. Ongoing technical development will continue under the Coalition for Secure AI, known as CoSAI. The backing consortium includes AMD and Intel for hardware attestation roots, Microsoft and its Azure confidential computing stack, and TII, based in the UAE. ## The problem TRACE is solving As organizations deploy increasingly autonomous AI agents to handle sensitive workloads across diverse infrastructure, a fundamental trust problem emerges. How does a regulated bank verify that its AI model ran inside a secure enclave on a specific cloud provider? How does a healthcare company prove to auditors that patient data was processed according to policy? How does any organization demonstrate compliance when the workload ran on someone else’s hardware? TRACE flips the model. Because the attestation originates from the hardware itself, the resulting Trust Records can be verified by any third party with access to the specification. The cryptographic verification can confirm how something ran without revealing what it processed. **Disclosure:** This article was edited by Editorial Team. For more information on how we create and review content, see our [Editorial Policy](https://cryptobriefing.com/editorial-policy/).