OPAQUE Releases Version 3.0, Launches First Verifiable Governance MCP Extension
2026-06-24 09:29
Favorite

en.Wedoany.com Reported - OPAQUE, an enterprise AI trust verification platform, recently released version 3.0. The platform provides verifiable trust mechanisms for AI agents through open governance standards, cryptographic identity, and confidential execution. Enterprises can prove what AI actually does, which components are running, and where, without relying on promises. OPAQUE 3.0 includes two open-source milestones: the release of the Agent Manifest standard built on the Agent Governance Toolkit (AGT), and the launch of Confidential MCP, the first implementation of the Model Context Protocol (MCP) to achieve verifiable governance through confidential computing.

The core features of OPAQUE 3.0 are all based on open standards. The Agent Governance Toolkit (AGT), created by Imran Siddique during his time at Microsoft, was released as an open-source framework and attracted approximately 4,100 GitHub stars in its first six weeks. This toolkit is used to define AI agent behavior permissions, data access scope, and governance policies. Agent Manifest extends AGT to runtime integrity verification, cryptographically verifying the agent's identity, resource access permissions, approving parties, and applicable governance policies. Confidential MCP runs within the OPAQUE confidential runtime, enforcing governance policies in hardware and generating independently verifiable evidence for each agent tool call.

The platform can run on CPU-based confidential computing environments from Intel, AMD, and NVIDIA, and supports GPU integration in NVIDIA confidential computing. OPAQUE 3.0 can be deployed across public cloud, private cloud, and on-premises environments. The Technology Innovation Institute (TII) of the United Arab Emirates, as a founding partner, applies the standards to sovereign AI deployments, and both parties are integrating TII's post-quantum cryptography into the platform to address the risk that future quantum-era attackers could break existing encryption. OPAQUE 3.0 is currently collaborating with multiple financial institutions, global enterprise software companies, and national sovereign AI construction projects.

OPAQUE was founded at the RISELab (now Sky Compute Lab) of the University of California, Berkeley. The founding team includes Dr. Ion Stoica (co-founder of Databricks, co-director of Sky Compute Lab at UC Berkeley), Dr. Raluca Ada Popa (ACM Grace Hopper Award winner, senior staff research scientist at Google DeepMind), and Rishabh Poddar (CTO). Imran Siddique serves as Chief Platform Officer. OPAQUE 3.0 made its debut at the Confidential Computing Summit and will be generally available in July 2026. Agent Manifest and Confidential MCP are open on github.com. The platform will also launch the agentrust-io Fellowship sponsorship program, with applications opening in August.

This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com