The rapid proliferation of artificial intelligence within the enterprise landscape has created a significant visibility gap, as organizations struggle to identify and govern the autonomous agents operating within their networks. In response to this growing security challenge, the Linux Foundation announced its intent on June 23 to launch the Agent Name Service (ANS), a new open-standard infrastructure designed to provide a verifiable identity layer for AI agents. Developed by GoDaddy in collaboration with industry giants including Cisco, Salesforce, Cloudflare, and Infoblox, ANS seeks to leverage the existing architecture of the Domain Name System (DNS) to bring accountability and transparency to the "agentic internet."
The move comes at a critical juncture for corporate cybersecurity. According to a recent survey conducted by the Cloud Security Alliance (CSA) and commissioned by Token Security, the scale of "shadow AI" is reaching a breaking point. The study, which polled 418 IT and security professionals, revealed that 82% of organizations have discovered unknown AI agents running within their infrastructure. This lack of oversight has tangible consequences: 65% of respondents reported experiencing an agent-related security incident in the past year. Of those breaches, 61% resulted in sensitive data exposure, while 35% led to direct financial losses. By anchoring AI identity to the 40-year-old DNS protocol, the ANS initiative aims to transform these "black box" entities into discoverable, attributable, and verifiable assets.
The Architecture of Trust: How ANS Utilizes DNS and PKI
The Agent Name Service is built on the premise that a new identity system should not require a proprietary namespace or entirely new hardware. Instead, it utilizes the existing naming layer that routes global internet traffic. When an enterprise deploys an AI agent, it publishes the agent’s record under a domain it already owns. This ensures that any third party interacting with the agent can verify who the agent works for by performing a standard DNS lookup.
The technical framework of ANS relies on a sophisticated implementation of Public Key Infrastructure (PKI). Each agent record is attached to two distinct certificates. The first is a standard server certificate issued by a public authority, similar to those used to secure websites. The second is a specialized identity certificate from a private authority that binds the agent to a specific version of its own code. This dual-layer approach allows for varying levels of assurance. For instance, organizations can opt for Organization Validation (OV) or Extended Validation (EV) certificates—the same high-security standards used by banks and major retailers—to prove that a legitimate legal entity stands behind the agent.
One of the most innovative aspects of ANS is its integration of a cryptographically sealed, append-only log. Inspired by the Certificate Transparency (CT) model used in web security, this log records every registration, version update, and revocation. Because the log is immutable, it prevents malicious actors from altering an agent’s history to hide fraudulent activity. For incident responders, this provides a reliable timeline to reconstruct what an agent was and how its permissions changed over time.
A History of Development: From OWASP to the Linux Foundation
The journey of ANS did not begin in a corporate vacuum. The service originated within the Open Worldwide Application Security Project (OWASP) before being refined by GoDaddy’s engineering team. It is currently a draft at the Internet Engineering Task Force (IETF), co-authored by researchers from OWASP, Cisco, and Scott Courtney, GoDaddy’s Vice President of Engineering. Courtney, a 16-year veteran of Verisign, brought deep expertise in DNS infrastructure to the project, ensuring that ANS could scale using the same systems that currently handle over 100 million active SSL and TLS certificates.
GoDaddy’s decision to hand the standard over to a neutral foundation like the Linux Foundation was a strategic move to ensure broad industry adoption. Jared Sine, Chief Strategy and Legal Officer at GoDaddy, emphasized that an identity layer only functions effectively if it is universally recognized. "You can’t really have an agentic internet if you don’t have identity for agents," Sine stated. He noted that while GoDaddy built and tested the system in production, a proprietary model would have failed to gain the necessary trust from competitors and the wider tech ecosystem.
The "intent to launch" phase is currently underway, with partners building working implementations to prove the specification’s viability. A notable live deployment involves Salesforce and its MuleSoft division, where agents are already being registered in ANS through DNS in real-time. This practical application demonstrates that the service is ready for enterprise-scale integration rather than being a distant roadmap item.
Distinguishing Internal Governance from External Portability
As AI adoption grows, major cloud providers like Microsoft, Google, and Amazon Web Services (AWS) have introduced their own identity models, such as Microsoft Entra Agent ID. However, the ANS initiative addresses a different problem set than these internal governance tools. While Entra and similar systems manage how a company’s agents behave within its own "walled garden," they do not provide a mechanism for verifying agents that cross organizational boundaries.
Consider a B2B scenario where a supplier’s procurement agent attempts to access a partner’s inventory system. Without a cross-platform identity standard, the receiving system has no reliable way to verify the agent’s origin or permissions. ANS fills this gap by providing a portable identity. Because every company maintains its own DNS, an agent can carry its verifiable credentials from one environment to another. This interoperability is designed to complement existing protocols like Agent2Agent and the Model Context Protocol, rather than replacing them.
Versioning and the Trust Index: Solving the "Stale Certificate" Problem
A significant weakness in traditional web security is that a TLS certificate only proves domain ownership; it says nothing about the code running on the server. If an e-commerce agent is compromised and begins manipulating prices, a standard certificate would remain valid, allowing the fraud to continue undetected. ANS solves this by tying identity to specific versions. When an agent’s code is updated or its model is changed, the old identity becomes invalid, necessitating a new registration.
To further enhance security, the ANS specification includes a "Trust Index." GoDaddy is currently developing this companion piece with plans to open-source it in the near future. The Trust Index will allow organizations to assign scores to agents based on various signals, such as the age of the domain, the class of certificate held (OV vs. EV), and historical performance. This allows security administrators to set strict boundary policies. For example, a company could decide to only admit external agents that possess an EV certificate and a Trust Index score above a certain threshold, turning agent admission into a policy-driven decision rather than a manual security hurdle.
Industry Implications and the Path to Adoption
The launch of ANS has significant implications for the future of autonomous commerce and digital labor. By providing a standard for "who" an agent is and "what" it is permitted to do, the service paves the way for more complex AI-to-AI transactions. However, the success of the initiative depends entirely on the network effect. For ANS to become the "gold standard," it must be adopted not only by tech providers but also by NGOs, governments, and small-to-medium enterprises.
The transition to an agent-centric economy requires solving the "black box" problem. Currently, interacting with an unknown AI agent is often described as a high-risk gamble. ANS provides a framework to identify the owner of the "box" and verify the version of the software inside. While it cannot inherently prevent a model from hallucinating or making poor decisions, it provides the forensic and identity tools necessary to hold the human operators of those models accountable.
Critics of DNS-based security often point to vulnerabilities such as DNS hijacking and registrar compromise. The ANS team acknowledges these risks but argues that the use of DNSSEC (DNS Security Extensions) and the transparency of the append-only log mitigate these vectors. The specification was developed through OWASP threat modeling, which accounted for approximately thirty distinct security vectors.
Conclusion: Setting the Stage for an Agentic Internet
The Linux Foundation’s intent to launch the Agent Name Service represents a foundational shift in AI governance. By moving away from fragmented, proprietary identity silos and toward a unified, DNS-anchored standard, the industry is taking a necessary step toward securing the future of autonomous systems. As AI agents become more integrated into supply chains, financial services, and customer support, the ability to answer the basic questions of identity—who an agent represents and what code it is running—will be the difference between a secure digital economy and one plagued by "shadow" risks.
With the backing of major infrastructure players and a commitment to open-source stewardship, ANS is positioned to become the primary registry for the next generation of the internet. As Jared Sine observed, the goal is to move past the era of "pins and needles" security and toward a transparent environment where every agent has a verifiable name, a clear purpose, and a permanent record.
