Linux Foundation launches Agent Name Service to give AI agents verifiable identity

The Linux Foundation has launched Agent Name Service (ANS), a new open standard designed to give AI agents a verifiable identity and a common way to be discovered online. The announcement, dated 11 November 2026, landed with eleven organisations already contributing to or adopting the specification, including Cisco, Cloudflare, DigiCert, GoDaddy and Salesforce. The core idea is simple and ambitious at once: agents that transact and act autonomously across the internet need the digital equivalent of a passport, and that passport should not belong to any single vendor.

What does the standard actually specify?

ANS deliberately builds on the Domain Name System (DNS), the internet's existing addressing infrastructure, rather than inventing a separate naming layer. The specification covers four functions: how AI agents are named and registered, how their identities are published through DNS and verified, how identity records are monitored for unauthorised changes, and how agents authenticate one another when they interact.

The technical work now proceeds in the open under a technical steering committee. The specification consists of seven documents covering:

• Domain-anchored naming
• Registration and versioning
• DNS publication
• Public log auditing
• Integrity monitoring
• Agent-to-agent authentication

The choice to reuse DNS and certificate-based verification is a design bet, not a coincidence. Supporters argue these systems are already widely deployed, well understood, and tied to established internet governance and security practices, which lowers the barrier to adoption. The takeaway: ANS wants to look like plumbing the internet already trusts, not a new protocol it has to learn.

Who is backing it?

Eleven organisations have contributed to or adopted the standard so far. The named backers are Cisco, Cloudflare, DigiCert, DistributedApps.ai, GoDaddy, HOL, Infoblox, Radix, Salesforce, OWASP and Trustmodel.ai. Notably absent from the list, at least at launch, are the largest model developers and cloud providers — a gap the announcement does not address.

Some related services already consume ANS identities in production-adjacent settings. The AGNTCY Agent Directory Service, itself another Linux Foundation project, supports verification of ANS identities. ANS Finder implements the Agentic Resource Discovery specification and has passed registry conformance tests, according to the Linux Foundation.

The project's champions frame shared identity as a precondition for internet-scale deployment of autonomous agents, rather than a convenience. The argument: without a common trust model, agent discovery and authentication fragment into isolated, vendor-specific approaches.

Jim Zemlin, chief executive officer of the Linux Foundation, said the project addresses a basic issue for autonomous software operating across different networks and services. "Autonomous AI agents are transacting and operating across global systems, but they can't succeed without verifiable identity," Zemlin said.

"Under the neutral governance of the Linux Foundation, ANS provides a common foundation for trust that the entire industry can adopt, implement and improve together," he added.

Why DNS, and why now?

Backers repeatedly connected the standard to recent security failures involving agents. Cloudflare's chief technology officer, Dane Knecht, was explicit about the motivation.

"Recent high-profile security incidents involving AI agents have reinforced a fundamental requirement for the agentic web: organizations need to know who an agent represents before trusting it to act. ANS builds on DNS to help answer that question using infrastructure the internet already relies on, rather than creating another proprietary system. At Cloudflare, we believe open standards are essential to making the agentic web both trustworthy and interoperable," Knecht said.

DigiCert's Brian Trzupek, senior vice president of product management, framed the identity problem in terms of access control. "Before an organization gives an AI agent access to its systems or data, it needs to verify that agent's identity. Think of it as a passport for AI agents, which is a concept ANS helps advance across platforms and organizational boundaries by bringing together DNS and certificate-based identity. DigiCert supports this open approach as part of a broader enterprise imperative of knowing which agents are operating, controlling what they can do, and having the ability to revoke their access when trust changes," Trzupek said.

That last clause — revocation — is worth watching. A registry that cannot reliably strip an agent's identity after trust breaks down would be a directory, not a control. The seven specification documents include integrity monitoring and public log auditing, but the announcement offers no detail on how revocation works in practice or how quickly tampered records propagate. In short: the promise is verifiable identity; the enforcement mechanics remain to be seen.

Does neutral governance matter here?

Several backers emphasised the decision to host the specification at the Linux Foundation rather than under a single commercial sponsor. The governance model is meant to reassure users that the naming and verification system stays open to multiple participants.

Ken Huang, chief executive officer of DistributedApps.ai, disclosed his company's role in the project's origins. "I led the initial development of ANS, and I am very glad to see the momentum and adoption that many organizations have built around it. By hosting ANS under neutral governance, the Linux Foundation gives the whole ecosystem a trusted home for agent registration, discovery, and security governance. I look forward to joining the technical steering committee and contributing further as the specification evolves in the open," Huang said.

Jared Sine, chief strategy and legal officer at GoDaddy, tied the standard to broader questions of open internet infrastructure. "Agent identity is essential to building trust in the agentic open internet, and open standards lay the foundation for ensuring it stays open," Sine said. He added that adoption across a range of projects shows "how agents can be discovered and verified beyond any single proprietary platform," calling the momentum "an exciting tailwind for bringing the agentic open internet to life."

Salesforce connected ANS to cross-platform governance for its own agent ecosystem. "True agentic transformation and AI scale require an open ecosystem where agents can be discovered, verified, and trusted across organizational boundaries. ANS will help set a global framework for agent identity to help ensure security and traceability. By aligning on open standards like ANS, Salesforce's Agent Fabric gives organizations deeper governance across diverse platforms, allowing developers the freedom to build anywhere while meeting enterprise security standards," said Andrew Comstock, senior vice president and general manager of MuleSoft and Agent Fabric at Salesforce.

What is still unproven?

TrustModel.ai offered the most concrete picture of how ANS might slot into layered trust systems. "Identity is the first question in agent trust, not the last. Before you can score, audit, or govern an AI agent, you have to know exactly which agent you're dealing with, and that answer can't belong to any one vendor. Neutral governance under the Linux Foundation makes ANS a shared trust anchor the whole industry can build on. At TrustModel.ai, we're anchoring agent TrustScores to ANS identities," said Karl Mehta, founder and chairman of TrustModel.ai.

That positioning is revealing. ANS supplies the identity layer; scoring, auditing and revocation sit on top. Each of those upper layers remains vendor territory, which means the "shared trust anchor" solves only the first link in the chain.

The specification now moves forward under its technical steering committee, with Huang among its contributors. Its early adopters already include interoperable tooling — the AGNTCY directory and the conformance-tested ANS Finder — and at least one vendor, TrustModel.ai, wiring ANS identities into its products. Whether the eleven-organisation roster grows to include the major AI platforms, or whether those platforms push rival identity schemes of their own, will largely determine whether ANS becomes infrastructure or another well-governed specification that the biggest players politely ignore.