The Linux Foundation plans to launch Agent Name Service (ANS), an open standard built on DNS to provide trusted identity, verification, and discovery
The Linux Foundation has formally announced plans to launch the Agent Name Service (ANS), a system built on the existing Domain Name System (DNS) infrastructure to establish an open identity verification and discovery mechanism for AI agents operating on the internet. This move directly addresses critical challenges faced by AI agents as they increasingly enter enterprise production environments, including identity confirmation, permission management, governance, and interoperability when executing tasks across organizations and platforms. ANS does not create a separate proprietary lookup network or a centralized registration system maintained by a single vendor; instead, it links AI agent identities to the existing DNS infrastructure, ensuring that system administrators, developers, and users can verify which organization an agent represents, what permissions it has obtained, and whether its code and operational history remain authentic and unaltered[1][2].
Previously, AI systems were mostly limited to responding to requests within a single service, but enterprises are now evaluating whether AI agents can call external tools, interact with other agents, or even execute partial processes on behalf of organizations. When agents operate across different services, relying solely on platform internal accounts or general API credentials makes it difficult to handle issues related to agent source, version, and trust boundaries. ANS, as an agent name resolution, registration, and transparency log mechanism, can resolve agent names into verifiable identity data with version information, and combine certificates with append-only, non-overwrite transparency logs to enhance overall transparency and security[1][2].
The core technical architecture of ANS is built on the existing DNS foundation, but its functionality extends to agent name resolution and identity verification. Traditional DNS resolves domain names into network addresses, whereas ANS resolves agent names into verifiable identity data with version information. This mechanism includes three key elements: agent name resolution, registration mechanism, and transparency log. The transparency log adopts an "append-only, non-overwrite" principle to ensure historical records cannot be tampered with, and is combined with digital certificates to strengthen verification intensity[1][2].
The ANS framework explicitly supports Decentralized Identifiers (DIDs) and Legal Entity Identifiers (LEIs), enabling organizations to integrate existing identity systems into a unified verification model for cross-system interoperability. Currently, ANS is an open standards initiative under development, with relevant technical documents appearing in IETF Internet-Draft; however, this status indicates a working draft, and ANS has not yet achieved formal IETF standard status. The project is seeking participation from enterprises, AI developers, infrastructure providers, and security researchers, with subsequent technical repositories and contribution methods to be announced through the Agent Name Service GitHub organization[1][2].
The launch of ANS will have broad implications for enterprise IT environments, AI development ecosystems, and infrastructure providers. First, enterprises introducing AI agents into production environments will gain a more reliable identity verification mechanism, reducing security risks caused by unclear agent identities or mismatched permissions. Second, AI developers can standardize agent identities through ANS, improving interoperability and security in agent interactions and reducing fragmentation. Additionally, infrastructure providers will need to adjust DNS architectures to support ANS functionality and strengthen verification processes in conjunction with transparency log mechanisms[1][2].
The standardization process of ANS may also impact existing identity management systems. By supporting DIDs and LEIs, ANS can integrate decentralized and traditional legal entity identities, providing a more flexible identity model for cross-organizational collaboration. However, since ANS is currently still an IETF working draft and not yet a formal standard, its final technical details and implementation timeline remain subject to change. If ANS fails to achieve widespread adoption, enterprises may face confusion from multiple identity verification standards coexisting, increasing integration costs and security risks[1][2].
In light of ANS not yet being a formal standard, enterprises and developers should adopt the following protective measures: First, before introducing AI agents, establish internal agent identity verification mechanisms to ensure agent source, version, and permissions are clearly verifiable. Second, continuously monitor the progress of ANS technical documents at IETF and participate in contribution discussions within the GitHub organization to stay informed about standardization directions. Additionally, enterprises should evaluate the scalability of their existing DNS architectures and prepare hardware and software resources needed to support ANS functionality[1][2].
Security researchers should focus on monitoring the implementation details of the transparency log mechanism to ensure its "append-only, non-overwrite" principle is not compromised and verify the strength of digital certificate verification. For enterprises that have already deployed AI agents, it is recommended to regularly review agent identities and permission configurations to avoid excessive permission grants or identity confusion. Finally, enterprises should establish approval processes for agent identity changes, ensuring all changes are recorded in the transparency log and can be traced back to the original operator[1][2].