On July 9, 2026, at the AI for Good Global Summit in Geneva, the International Telecommunication Union announced the creation of a new Focus Group dedicated to one of the most fundamental challenges of the agentic AI era: how to establish trust and identity when it is no longer only humans who act, but autonomous agents acting on their behalf.

The new Focus Group on Trust and Identity for Humans and Agentic AI (FG-TIDA) will work to define trust management mechanisms and an interoperable digital identity infrastructure, designed for both humans and agentic AI systems. Its goal is to foster the development of safe and reliable digital ecosystems, in which humans and AI agents can interact and collaborate securely. (Source: ITU-T FG-TIDA)

We Have Already Lived Through This

Thirty years ago, the global music industry shifted into the digital age. It was the far west. Piracy was exploding, the identities of works and rights holders were unclear or simply ignored, transactions happened without traceability, without common rules, without trust.

Then the industry organized itself. Not by holding back technology, but by giving it rules. Universal identifiers for works. Standardized metadata. Interoperability protocols between platforms, rights holders, distributors and consumers. Machine-readable contracts. Clarity created trust. Trust created value. A global ecosystem of high added value — for creators, networks, technology players and audiences worldwide — was born from that collective discipline.

Thirty years later, in the age of agentic AI, we are at the same crossroads. And we have a choice: let the digital far west make a full comeback, or build right now the digital roads of trust that will make this space safe, transparent and dynamic for all its participants.

That is precisely the ambition of FG-TIDA.

A Structural Turning Point in the History of the Digital

AI agents are capable of reasoning, planning and executing tasks autonomously. They are changing the very nature of digital identity. For thirty years, the question was simple: one human, one account, one authentication. Trust systems were built around that assumption.

That assumption is now obsolete.

An agent can place orders, access databases, interact with other agents, make decisions in sequence — without direct human intervention at every step. It acts on behalf of someone. But how do we know, at every moment, on whose behalf? With what authority? Under what conditions? And with what guarantee that its actions remain aligned with the intentions of the person it represents?

The risk is not abstract. An AI agent can impersonate a person or an organisation. It can make unauthorized decisions. In domains as sensitive as financial transactions or critical infrastructure, the consequences can be severe and difficult to reverse.

These questions are being asked today. They call for shared answers, at the international level.

Why the ITU, and Why Now

The International Telecommunication Union is the United Nations agency for digital technologies. It produces the standards that make global networks function. When the ITU takes ownership of a subject, it is because that subject has become critical enough and mature enough to warrant a coordinated response at the global scale.

FG-TIDA is attached to Study Group 17, the ITU's group of experts dedicated to the security of telecommunication standards — one of the most influential normative bodies in the area of cybersecurity and digital infrastructure.

"The future of AI depends on trust," said Doreen Bogdan-Martin, ITU Secretary-General. "As AI becomes more autonomous, we need to work together — industry, governments, academia and civil society — to ensure the highest level of trust in AI systems."

This conviction is not new for those who have been building responsible digital infrastructures for years. Reconciling ethics and the digital, creating roads of trust between humans and machines, playing the card of technological transparency: these are principles that long predate the emergence of agentic AI. FG-TIDA finally gives them a global normative dimension.

What the Focus Group Will Build

The mandate is precise. This is not a general-purpose AI reflection group, nor a governance forum in the broad sense. The objective is technical and operational: to produce shared foundations so that humans and AI agents can interact in a safe, traceable and interoperable way.

The digital publishing sector perfectly illustrates the complexity of these challenges. Citing a work, indexing it, using it to train a model, generating derivative content, contracting rights — each of these actions involves multiple actors, human and machine, whose identities, authorizations and responsibilities must be clearly established at every step. This is a use case for illustrating what FG-TIDA could address, among others, to anchor its frameworks in the reality of the digital creative economy.

The group will produce six categories of deliverables.

Use case analysis and requirements. Identifying, sector by sector, the real situations in which humans and AI agents interact, and deriving from them the technical requirements for interoperable identity and trust management.

Architectures for identity, trust, agent discovery and interoperability. Defining architectural approaches covering agentic identity stacks, cross-border trust management, and real-time control mechanisms across heterogeneous environments.

Trust frameworks and lifecycle management. Modelling how trust is established, continuously assessed, maintained and revoked, integrating human oversight and behavioural trust indicators for agents.

Machine-readable trust metadata. Specifying standard formats to represent and exchange trust and identity information in an automatable way across heterogeneous systems.

Guidelines, metrics and standardisation recommendations. Producing evaluation criteria for the reliability and interoperability of agents, and formulating recommendations for forthcoming formal standardisation work.

Coordination. Maintaining active liaison with ITU-T study groups, external standardisation bodies, academia and industry, and delivering a comprehensive final report to Study Group 17.

The Central Challenge: Delegating Without Losing Control

Trust can be delegated. Identity cannot.

Take the case of content platforms. Every day, AI agents from every continent knock at the door of digital infrastructures. Some come to index. Others to cite, mine, extract for model training. Others still to trigger processes, act on rights. These agents have precise objectives. They have principals. They have — or should have — a verifiable identity.

Protocols already exist to lay down the rules of access: digital roads, machine-readable rights policies, governance frameworks defining what an agent can do, under what conditions, with what limits. That is a real step forward. But it hits a wall.

A digital contract can only be formed if both parties are identified. An agreement on rights can only produce legal effects if we know who agreed, on whose behalf, and with what capacity to commit. A transaction can only be industrialized if the identity of the agent that triggers it is known, verifiable, and tied to a clear chain of responsibility.

That is the missing link. The rules are there. The roads are laid. But if the agent knocking at the door does not declare itself — its identity, its capabilities, its limits, its principal — then the contract remains a dead letter. Trust cannot be established in agentic anonymity.

This is precisely what FG-TIDA must resolve: not merely defining what agents are allowed to do, but guaranteeing that we know who they are before they act.

"We need common international foundations that establish who these agents are, under what conditions they can be trusted, and how people will remain in control of their actions," said Debora Comparin, co-chair of the Focus Group. "That is the challenge this Working Group was created to meet."

"Identity tells us who is acting. Trustworthiness tells us how that actor is likely to behave. Bringing these two dimensions together creates the shared foundation necessary for AI systems that are interoperable, accountable and trustworthy at global scale," added co-chair Amir Banifatemi.

Clarity as a Condition for Trust

"The consensus we build during the development of ITU standards creates the clarity we need for new innovation ecosystems to develop," said Seizo Onoe, Director of the ITU Telecommunication Standardisation Bureau. "This Focus Group is an open platform, designed to help everyone move forward with confidence and certainty."

Some platforms have already undertaken this work with clarity, transparency and an educational approach. In the digital platform sector, trust infrastructures have been built layer by layer so that every actor in the value chain knows exactly who they are dealing with and under what conditions they are interacting. Machine-readable digital rights protocols define what an agent can do with a piece of content: index it, cite it, use it to train a model — or not. Encoded, verifiable and enforceable rights management policies set the conditions for a digital contract before the transaction is even triggered. This is not a vision. It is operational, already deployed and already auditable.

This is what FG-TIDA intends to generalize at global scale, before the agentic far west takes permanent hold.

An Open Group, a Global Ambition

The Focus Group is open to interested experts. It brings together governments, telecommunications players, cybersecurity and digital identity specialists, legal experts and representatives from highly diverse sectors — reflecting the cross-cutting dimension of the subject.

"We are moving strategically and rapidly, but deliberately taking the time to lay the foundations properly," said Arnaud Taddei, President of Study Group 17. "We know the direction and, in broad terms, what we want to build. We are currently bringing together the right people and the right structure."

Its first meeting will be held in Paris in November 2026. The second in Geneva in January 2027.

_____________________________

Text written by Alexandre Leforestier, CEO of Panodyssey and founder of the European consortium CREA Trust AI, during his participation in the AI for Good Global Summit events in Geneva in July 2026.

Official source and information on joining the group: ITU-T Focus Group on Trust and Identity for Humans and Agentic AI, TIDA