Entrust launches AI agent trust program as ITU develops identity frameworks

Entrust has launched the Agentic AI Trust Accelerator to help enterprises move autonomous AI agents from pilot projects into production environments. The program focuses on the identity, authorization, and cryptographic controls that will determine who an agent represents, what it is allowed to do, and how its actions can be verified.

In a separate initiative around agentic AI, the International Telecommunication Union (ITU) has created a new focus group on trust and identity to develop frameworks for digital identity and assess if AI agent behavior remains accountable throughout the agent lifecycle.

Entrust proposes a “trust plane” for autonomous AI

The Agentic AI Trust Accelerator program is open to selected enterprise customers, financial institutions, cloud and SaaS providers, systems integrators and other technology partners.

These participants will identify practical use cases, validate reference architectures and develop controls for AI agents operating across applications, business processes and organizational boundaries.

“Enterprises need to be able to trust autonomous actions across business processes, partners, and systems. Whether organizations are experimenting with AI agents, deploying initial use cases, or preparing for broader adoption, they need a trust foundation that can scale with them,” says Anudeep Parhar, COO for digital infrastructure at Entrust.

The company describes the proposed initiative as a “trust plane” for autonomous AI. The concept is organized around four areas: identity, authorization, cryptographic trust, and accountability.

Entrust says both the human and the AI agent should have verifiable identities to allow an action to be traced back to a confirmed person and a unique agent. Entrust proposes authorization through checking permissions in real time so that an agent can act within approved policies and roles.

These AI agents often need credentials to authenticate themselves to applications and APIs. Entrust plans to protect the keys, certificates, and signing capabilities used in those interactions. Entrust also wants to create cryptographically verifiable records of agent actions that could be reviewed by internal risk teams, customers, or even regulators.

The accelerator will explore how those technologies can be applied when autonomous agents operate across multiple applications, partners, and enterprise systems.

“The goal is to control AI agents, but to operate with confidence that they are appropriately authorized, governed, and acting within established business, security, and regulatory boundaries,” says Emanuel Figueroa, senior research analyst for Identity Security, Worldwide at IDC.

ITU forms a focus group to work on agent identity and trust standards

There is a growing risk that AI agents will impersonate people or organizations to commit unauthorized actions and operate across connected systems without sufficient human oversight.

“As AI becomes more autonomous, we need to work together across industry, governments, academia and civil society to ensure the greatest possible confidence in AI systems,” says Doreen Bogdan-Martin, secretary general at ITU.

A key aspect of the focus group is the differentiation between identity and trustworthiness. Identity establishes which person, organization, or AI agent is acting.

For example, it answers questions such as which agent initiated a transaction, which organization deployed it, and which human principal it represents. But trustworthiness addresses a different question of whether that actor can be expected to behave within approved business and regulatory boundaries.

The group also plans to develop reference architectures to cover identity, trust, agent discovery, and interoperability. They will also examine interoperability mechanisms for digital identities and credentials.

In addition, the ITU wants to develop security criteria and benchmarks for the continuous assessment of AI agents. This will allow organizations to evaluate if an AI agent is still operating within its approved role and if intervention is required.

“Before that future becomes reality, we need common international foundations that establish who these agents are, when they can be trusted, and how people will remain in control. That is the challenge this Focus Group has been created to address,” says Debora Comparin, co-chairman of the Focus Group.

The focus group will report to ITU-T Study Group 17. Its first meeting is scheduled for November 2026 in Paris, followed by a second meeting in Geneva in January 2027.

Article Topics

AI agents  |  authentication  |  digital identity  |  digital trust  |  Entrust  |  ITU standards