The Linux Foundation plans to launch ANS, a standard for AI agent identity built on DNS

The Linux Foundation has announced its plans to launch the Agent Name Service, an open standard for identifying artificial intelligence agents. The initiative leverages the domain name system, the infrastructure that already underpins the web, allowing for the verification of an agent’s identity, permissions, and code integrity. The project has garnered support from several internet and software leaders, including Cloudflare, Cisco, Salesforce, and GoDaddy.

As AI agents transition from experimental projects to production systems, operating across various applications, platforms, and services, the need for a robust identity framework becomes evident. When an organization permits an agent to interact with other systems, it is essential to ascertain the agent’s ownership, its authorized capabilities, and whether its code has remained uncompromised. Without such insights, the agent risks being perceived as an opaque entity, leaving stakeholders unable to assess its legitimacy.

The Agent Name Service aims to bridge this gap. Rather than creating a proprietary registry, the Linux Foundation ties agent identities to the domain name system, a global directory responsible for resolving web addresses and already handling over 100 million requests per second. Organizations will be able to identify their agents using the domain names they own, eliminating the need for intermediaries or reserved naming spaces. The project is currently in the intention-to-launch phase, with its technical repositories open for contributions.

An Identity Link for Production

The need for such an initiative is far from theoretical. According to the World Economic Forum, 82% of leaders anticipate adopting AI agents within the next one to three years, even as the mechanisms for assessing and managing these agents remain uncertain. Identity is among the top concerns, alongside authentication, governance, and interoperability, that IT leaders face when these agents move beyond isolated deployments.

For CIOs and CISOs, the challenge aligns with existing issues surrounding machine identities, particularly with service accounts and API keys. An autonomous agent operating without human validation at every step places a greater burden of responsibility on its operator, necessitating precise tracking of its actions and permissions. A verifiable identity layer addresses this requirement, provided it encompasses the entire lifecycle of the agent—from registration to revocation.

Verifying Who Acts and With What Rights

The service will answer three critical questions when an agent presents itself to a system. First, it will identify the organization the agent represents, thus determining the responsible party behind it. Second, it will clarify the permissions assigned to the agent, defining the scope of its potential actions. Finally, it will ensure that the agent’s code and operational history have not changed since their initial registration.

This verification is particularly crucial for interactions between agents owned by different parties, where no prior trust relationship exists. An agent requesting assistance from another or a service receiving a query from an unknown agent can then confirm the legitimacy of the requester before proceeding. The Linux Foundation highlights this point as a solution to the risk of uncontrolled agents operating without oversight—an ongoing concern for security teams.

Leveraging DNS and Existing Identifiers

The technical choice here involves utilizing an established infrastructure rather than creating a new one. The DNS has secured web navigation for four decades, and extending its use to agents helps prevent fragmentation that could arise from competing systems. The framework accommodates decentralized identifiers and legal entity identifiers, enabling organizations to link their agents to identity references they already employ.

However, this remains a declaration of intent, and its impact will depend on actual adoption by providers and developers. The governance provided by the Linux Foundation aims to be neutral, ensuring that no single player controls the directory of agents. Nevertheless, the list of initial supporters, primarily North American, indicates that the future management of this identity layer may lean towards major internet players—a dependency that European organizations will need to evaluate before fully embracing it.

Scroll to Top