New framework addresses a critical gap in cloud and AI security by moving attestation beyond startup verification to continuous runtime assurance

NEW YORK: MITRE's latest framework signals a fundamental shift in how organizations will establish trust in AI, cloud and critical infrastructure. Developed with founding co-authors Fr0ntierX and Invary through the Confidential Computing Layered Attestation Working Group, the Framework for Continuous Remote Attestation is the industry's first open, implementation-neutral architecture for continuously verifying the integrity of systems while they are running, rather than only when they start.

Fr0ntierX contributed real-world deployment experience from production confidential computing environments, helping shape a framework designed to address how organizations can continuously trust the infrastructure their autonomous systems depend on.

"Establishing trust in the infrastructure that underpins AI isn't a problem any one company can solve alone," said Jonathan Begg, CEO of Fr0ntierX. "It requires an open architecture that works across cloud providers, Trusted Execution Environment (TEE) platforms and compliance regimes. That's why we're working with MITRE and Invary to help define a common framework for continuous remote attestation."

Fr0ntierX's Polaris platform runs sensitive workloads inside hardware-protected environments where data stays encrypted even while it is being processed, including from the cloud provider itself. That deployment experience directly informed key sections of the framework.

The new standard builds on the IETF remote attestation architecture with four key advances: continuous verification across hardware, software, and runtime environments; trust models tailored to the needs of cloud providers, enterprises, application operators, and end users; attestation controls for AI workloads including model integrity, RAG data sources, and GPU-based Trusted Execution Environments; and built-in support for post-quantum cryptography.

"Confidential computing gives organizations a trusted foundation for isolating sensitive workloads, but isolation alone doesn't answer the full integrity question," said Jens Albers, CTO of Fr0ntierX. "This framework defines how hardware-rooted attestation, workload identity, runtime evidence, and attested communications work together continuously in production, not just periodically."

The public review is open through August 31, 2026, with feedback invited from the confidential computing community. The full framework and supporting documents are available on the MITRE website: https://www.mitre.org/news-insights/publication/framework-continuous-remote-attestation. Additional information about Fr0ntierX is available at https://www.fr0ntierx.com.

Fonte: Business Wire