New KYA service aims to give AI agents verifiable identities

Solving the AI accountability gap

As artificial intelligence moves toward 'agentic' behavior, software is increasingly making independent decisions. These agents can manage digital services, access business information, and even facilitate payments without direct human intervention. While this increases productivity, it creates a significant trust gap for organizations.

Without a way to verify who is behind an AI action, companies struggle to distinguish authorized activity from unauthorized systems. The KYA service, launched by Zetrix AI and the China Academy of Information and Communications Technology (CAICT), aims to solve this by providing a verifiable identity for every agent. It moves the focus from what an AI can do to who is responsible for what it does.

The service is designed to help businesses and governments operate with confidence. By establishing a clear link between an AI agent and its human or corporate owner, the framework ensures that every action can be traced back to a responsible party. This is a necessary step for AI to be adopted at scale in sensitive environments.

How verifiable credentials work

The KYA service functions by assigning each participating AI agent a digital identity linked to an authenticated owner. Unlike standard digital identities that only verify humans, this system extends those same principles to the software agents themselves. It uses W3C Verifiable Credentials and AvatarLabs' Agentic Wallet technology to manage these digital credentials.

Owners of these agents can set explicit boundaries on what the AI is allowed to do. For example, a business could authorize an agent to pay approved suppliers but set a specific spending limit. If a transaction exceeds that amount, the system would require a human to provide approval before the payment can proceed.

The framework also handles complex scenarios like delegated authority. If an AI agent is assigned a task by a human and then delegates that task to another AI agent, the system ensures the second agent cannot exceed the original permissions. All activities are recorded as signed records, providing an auditable trail for security audits and oversight.

Testing across global borders

Because AI agents will likely operate in international markets, Zetrix AI and CAICT are launching a cross-border pilot program. This initiative will involve organizations from the Middle East, Africa, China, and the ASEAN region. The goal is to see how these identities can be recognized across different national jurisdictions.

A major challenge for this technology is respecting local regulatory requirements and data protection laws while maintaining a global network. The pilot will test how identity verification and permission enforcement work in real-world environments. It specifically looks at applications in cross-border trade, digital payments, and interactions between different autonomous agents.

The project aims to create a trusted network where agents can move across borders seamlessly. By testing these capabilities now, the partners hope to demonstrate that AI can be interoperable without sacrificing local control. This research will help determine how to balance global connectivity with regional sovereignty over identity systems.

Influencing global AI standards

The findings from the cross-border pilot are intended to have a significant impact on international policy. Zetrix AI and CAICT plan to submit their results to several major Standard Development Organizations (SDOs). These include the International Telecommunication Union (ITU) and the IEEE, which are currently discussing global standards for AI identity.

The submissions will include specific implementation findings and recommendations for delegated authority. By providing technical evidence from real-world use cases, the project seeks to help shape the rules for how the world handles agentic AI. This work is part of a larger effort to create a foundation for secure and responsible autonomous interactions.

The ultimate goal is to establish a common approach to trust and accountability for AI agents worldwide. As these systems become more deeply integrated into everyday digital activities, having a standardized way to verify authority will be essential. This initiative positions KYA as a foundational layer for the next era of AI adoption.