Linux Foundation launches Agent Name Service for AI agent identity
The Linux Foundation announced Agent Name Service (ANS), a new DNS-anchored open standard for AI agent identity, verification, and discovery, with launch partners Cloudflare, GoDaddy, Salesforce, Cisco, and Infoblox.
The Linux Foundation announced on June 23, 2026 the intent to launch the Agent Name Service (ANS), a new open standard built on top of DNS that gives AI agents a portable identity, verification, and discovery mechanism across the open web. The launch partners include Cloudflare, GoDaddy, Salesforce, Cisco, and Infoblox.
ANS arrives at the moment the agent identity problem has moved from a workshop concern to a board-level concern. World Economic Forum data cited in the announcement puts the adoption curve at 82 percent of executives planning to roll out AI agents within the next one to three years, against a backdrop of widely reported uncertainty about how to authenticate, govern, and audit autonomous systems that operate across enterprise boundaries. The same Forrester Identiverse 2026 recap that framed agent identity as the new IAM front and the AppViewX Agent Identity Security launch both made the case that portable identity is the missing primitive, and the Linux Foundation is now stepping into the open standards gap that those vendors have been working around. The wider enterprise agent governance debate, including the recent Pentagon Agent Network rollout, is all waiting on a portable identity layer to be a complete policy story.
The design choice to anchor ANS in DNS is the most consequential call in the framework. Rather than introducing a parallel lookup network, ANS ties agent identity to the same domain names an organization already owns, and uses the DNS infrastructure that already processes more than 100 million queries per second worldwide. The framework also supports decentralized identifiers (DIDs) and Legal Entity Identifiers (LEIs), so enterprises can fold their existing identity systems into a single verification model without re-issuing credentials. The intent is to let any organization identify the agents it operates, verify the permissions those agents carry, and check whether the agent code and operational history remain authentic and unchanged. The same properties that made DNS a successful federation layer for the public web, including decentralization, neutrality, and universal reach, are what the Linux Foundation is now offering for agents, and the choice to use DNS rather than a custom protocol is what gives the framework a real chance of multi-vendor adoption from day one.
The launch partners and what each one is bringing
The partner list is the other half of the story. Cloudflare contributes the lead-author research on ANS and the network footprint to serve agent identity queries at internet scale, with Dane Knecht framing the risk in the announcement as "shadow AI" , agents that proliferate without a neutral, trustworthy identity and discovery layer. GoDaddy brings the registrar side, the practical mechanics of mapping an agent identity to a domain an organization already owns, and the small-business and prosumer distribution channel that enterprise-only identity stacks typically miss. Salesforce brings the enterprise application side and the customer-facing agent surface, where agent identity is the difference between a trusted integration and an unauthenticated call. Together, Cloudflare and GoDaddy cover the public-facing resolution path from the network side and the registrar side, and Salesforce covers the application surface where agents actually meet customers and other agents.
Cisco contributes the network and security equipment side and is also working on complementary open standards at the IETF, with Nathan Jokel making the case that "every platform shift creates a choice between walled gardens and open ecosystems, and agentic AI is no exception." Infoblox brings DNS security and the policy layer that decides which agent identities are valid and which should be flagged or revoked. HOL, the human organization behind the original ANS research paper, is the convener and the intellectual anchor for the design. The Linux Foundation governance model, with its neutral IP stewardship and its track record on DNS, Kubernetes, and the PyTorch Foundation, is being offered as the home for an open alternative to the proprietary stacks that have been gaining traction across the enterprise vendor landscape over the last 18 months.
The launch is also a deliberate counter-move to the wave of proprietary agent identity stacks that have started to appear across the enterprise vendor landscape. Each of those stacks is a walled garden by design, and the vendors that ship them have an economic incentive to keep them walled. The Linux Foundation's bet is that the federation problem for agent identity is too important to be solved by a single vendor, and that the same model that worked for DNS, a neutral open standard with broad multi-vendor support, will work for agents. The new Agent Name Service GitHub organization is the public entry point, and the foundation is asking for participation from enterprises, AI developers, infrastructure providers, and security researchers as the technical and governance work ramps up over the rest of 2026. The first technical artifacts and contribution guides are expected to land in the third quarter, with the first stable specification targeted for early 2027.
The enterprise agent stack in 2026 is the test
The practical impact for an enterprise team rolling out agents in 2026 is that a portable identity layer is finally arriving. Today, every agent identity implementation is bespoke, every vendor rolls its own credential format, and there is no neutral way for an agent from vendor A to authenticate itself to vendor B without a custom integration. ANS, if it lands the way the Linux Foundation and its launch partners intend, is the equivalent of the DNS resolution step for agents: a globally consistent lookup that any system can call, with the verification result tied to a domain the agent operator already controls. For the enterprise agent stack, that means the difference between a federated agent ecosystem and a series of one-off vendor islands is now a standards-track conversation, not a custom build.
The procurement implication is also significant. Enterprise security teams that have been writing agent governance checklists and identity playbooks off the back of the Forrester Identiverse recap, the AppViewX launch, and the Pentagon Agent Network rollout now have a near-term candidate open standard to anchor those policies on. The full enterprise AI governance checklist covers action allowlists, logbooks, rollback paths, and accountability, and ANS is the first widely-supported candidate to fill the identity and discovery slot in that checklist. The full announcement and the partner list are on the Linux Foundation press release dated June 23, 2026, and the technical repositories will be published through the new Agent Name Service GitHub organization. The first 12 months of the standard will tell whether the open federation model wins or whether the proprietary stacks pull the agent ecosystem into closed gardens.
Weekly newsletter
Get a weekly summary of our most popular articles
Every week we send one email with a summary of the most popular articles on AIntelligenceHub so you can stay up-to-date on the latest AI trends and topics.
Comments
Every comment is reviewed before it appears on the site.
Related articles
Pentagon launches Agent Network for U.S. military command
The Pentagon has launched Agent Network, an agentic-AI tool that scans intelligence feeds and gives U.S. commanders targeting options within seconds, with Palantir and Lumbra as the lead contractors.
Nokia and Google Cloud ship six Gemini agents for telecom
Nokia and Google Cloud announced a six-agent Gemini stack for telecom network operations on June 22, two of which are already live, with the rest shipping in September 2026.
Meta open-sources Astryx, a React design system for AI agents
Meta open-sourced Astryx, a React design system with a Model Context Protocol server and CLI built for AI coding agents, MIT licensed and based on StyleX.