July 9, 2026, (Inside AI) — The International Telecommunication Union (ITU), the United Nations' digital tech agency, has launched a new initiative to build trust in AI agents. The announcement came Thursday at the AI for Good Summit in Geneva, amid rising concerns over accountability and human oversight of increasingly autonomous systems.
The ITU will establish a Focus Group tasked with creating frameworks to ensure AI agents remain identifiable, trustworthy, and subject to meaningful human control. The first meeting is set for Paris in November, with a second in Geneva in January.
AI agents are a new breed of AI designed to act independently on behalf of users, handling tasks from scheduling to complex business processes. While they promise productivity gains, risks include impersonation and unauthorized decisions, especially in sensitive areas like financial transactions and critical infrastructure.
"AI agents will soon negotiate, transact and make decisions on our behalf," said Focus Group Co-Chair Debora Comparin. She stressed the need for common international foundations to establish who the agents are and how and when they can be trusted.
The group will bring together technical, policy, and legal experts to tackle the fragmented regulatory landscape. This move echoes earlier ITU efforts, such as the AI for Good platform launched in 2017, but now zeroes in on the agentic layer that traditional AI governance has largely overlooked.
Unlike generative AI models that produce content, AI agents operate with a degree of autonomy, planning and executing multi-step tasks. This shift raises the stakes: a hallucinating chatbot might misinform, but an agent that books a flight or signs a contract without consent can cause immediate harm. The ITU's focus on identifiability and trust directly addresses these novel risks.
Competing viewpoints exist. Some industry players advocate for self-regulation, while others push for binding treaties. The ITU's approach, grounded in multi-stakeholder consensus, aims to bridge these divides. However, critics note that non-binding frameworks may lack teeth, especially when tech giants prioritize speed over safety.
Historical context is instructive. In 2023, the EU's AI Act set a precedent for risk-based regulation, but it didn't specifically address agentic systems. The ITU's initiative fills this gap at a global level, potentially influencing standards in regions without comprehensive AI laws. Yet, the absence of major AI powers like China and the U.S. from the initial working group could limit its impact.
The Focus Group will likely grapple with technical challenges, such as embedding verifiable credentials into agent architectures. It may also explore "human-in-the-loop" mandates for high-stakes decisions, a concept already debated in autonomous weapons discussions. The group's composition—spanning technical, policy, and legal experts—signals a holistic approach.
What's missing from the announcement is a timeline for concrete deliverables and enforcement mechanisms. Without clear milestones, the initiative risks becoming another talking shop. Moreover, the rapid pace of AI development could outstrip the group's work, as agents become more sophisticated and deeply embedded in digital ecosystems.
The ITU's effort is a welcome step, but its success hinges on broad adoption and adaptation to evolving AI capabilities. As AI agents move from hype to reality, the frameworks born from this Focus Group could either become the bedrock of global trust or a footnote in the history of AI governance.
