THEIA INSTITUTE · PUBLIC REVIEW

AI Agent & Cybersecurity Standards

What should an AI agent have to prove before it can act?

Theia Institute’s Standards Workspace is a public resource for evaluating AI agents and contributing to draft industry standards. Explore proposed safeguards, see what evidence could support them, and try practical scenarios before contributing your own.

Public discussion draft. Requirements and example test criteria are subject to review.

Contribute to the AI Assurance Initiative

From Draft Requirements to Public Input

AI assurance is the practice of gathering and evaluating evidence that an AI system manages risk and performs as intended in a defined setting. Policymakers, researchers, builders, and affected communities can use this workspace to ask what could go wrong, what should be tested, and what would count as sufficient evidence.

Define the use and who could be harmed.
State the proposed safeguard and how to test it.
Identify who can intervene and when the system should be reviewed again.

The initiative also welcomes pilot and draft-review collaborations and recommendations or expressions of interest in jointly signing Theia’s early 2027 AI policy letter.

Explore the Theia AI Assurance Initiative