IAB rejects ITU-T request for top-down AI agent standardization

The IAB's refusal to adopt a top-down approach ensures that AI agent protocols remain driven by engineering consensus rather than intergovernmental mandates. For the streaming industry, this means identity and authentication standards for AI-driven video workflows will likely emerge from established groups like WIMSE and OAUTH, which already handle cloud-scale security. This decentralized approach prevents a single regulatory body from bottlenecking technical innovation in AI-to-AI communication. As streaming platforms increasingly deploy autonomous agents for content discovery and metadata management, the industry must track the DAWN charter's progress to see which specific technical frameworks gain official IETF backing.

The IETF has multiple active working groups addressing AI agent identity and authentication, reflecting broad industry demand for interoperable standards. The WIMSE (Workload Identity in Multi System Environments) working group has been developing token-based identity frameworks for non-human workloads, including AI agents operating across cloud boundaries. In parallel, the DAWN (Decentralized Agent-to-Agent Networking) charter was submitted to the IETF in early 2025 to define protocols for autonomous agent discovery and communication, though it has not yet been approved as a full working group. The OAUTH working group continues to extend its authorization framework to cover machine-to-machine and agent-to-agent scenarios, which are directly relevant to streaming platforms deploying autonomous agents for content metadata and recommendation pipelines.

The institutional tension between the IETF and ITU-T over AI governance has intensified throughout 2025 and 2026. ITU-T Study Group 17 approved a preliminary framework for AI agent identity in March 2026, positioning it as a complement to existing internet standards. However, the IAB's liaison response makes clear that the IETF considers its consensus-based process the appropriate venue for protocol-level decisions. This mirrors a broader pattern: the IAB published a statement in November 2025 reaffirming that AI-related protocol work belongs in IETF working groups, pushing back against attempts by intergovernmental bodies to impose architectural requirements from above. For streaming technology vendors, the practical implication is that AI agent authentication standards will likely emerge from WIMSE and OAUTH specifications rather than ITU-T recommendations.

On the technical side, the agentproto effort represents one of several community-driven proposals for standardizing agent communication. A draft specification for agent-to-agent protocol interoperability was presented at IETF 122 in March 2025, drawing interest from participants working on multi-agent orchestration for media workflows. Meanwhile, the DMSC (Decentralized Messaging and Secure Communication) research group has explored how decentralized identity can support agent trust without centralized authorities. These parallel efforts underscore why the IAB insists on bottom-up development: multiple competing approaches need engineering validation before any single framework receives standardization backing, a process that top-down mandates would short-circuit.