On July 9, 2026, the International Telecommunication Union announced a new effort to develop international frameworks for the identity and trustworthiness of autonomous AI agents. The ITU's Focus Group on Trust and Identity for Humans and Agentic AI is expected to study common terminology, identity and trust architectures, agent discovery, credential interoperability, lifecycle models, security criteria, benchmarks, and a roadmap for future standards. This lands in a year already crowded with competing identity efforts: NIST's CAISI opened an AI Agent Standards Initiative in February, the Cloud Security Alliance published an Agentic Trust Framework, Microsoft shipped Entra Agent ID, and Identity Digital pushed a DNS-anchored proposal into the IETF. The ITU's move is different in kind — it's a UN specialized agency trying to set a floor across all of them, the way it once did for telecom numbering and spectrum. The stated goal is narrow but consequential: letting an institution — a bank, a platform, a government service — tell an authorized agent from an impersonator, and deciding which agents get to act on its behalf at all. As one analysis framed it, the initiative could help distinguish an authorized agent from an impersonator, and could also influence which agents institutions permit to operate. The harder problem isn't whether agents need credentials — nearly everyone building agent infrastructure now agrees they do. It's who issues those credentials, which organizations are obligated to recognize them, how much information a credential reveals about the agent or its owner, and who has the power to revoke it. A workable system needs multiple issuers, pseudonymous identities, selective disclosure, short-lived permissions, transparent revocation, and support for offline validation — otherwise an open technical standard can still calcify into a closed ecosystem if major platforms only accept credentials from a short list of approved issuers. That's the real fork in the road. Standards bodies like ITU, NIST, and IETF are producing technical specifications faster than anyone is answering the governance question underneath them: who gets to be a root of trust for a billion autonomous agents, and what happens to an agent — or its owner — when that trust is withdrawn. The Focus Group's charter puts "agent discovery" and "credential interoperability" on the same agenda item as "security criteria" and "lifecycle models," which suggests the ITU intends to treat identity and discoverability as one problem rather than two. Whether that produces an open, federated system or a small number of gatekeeping issuers is the question worth tracking over the next year, not whether agents get credentialed at all.