The Linux Foundation Moves to Launch DNS Identity Standard for AI Agents

Key Takeaways

• Agent Name Service would use existing DNS infrastructure to identify, verify, and discover AI agents across the internet.
• The federated approach is intended to reduce dependence on proprietary registries while supporting DIDs and LEIs.
• ANS could strengthen agent governance, but enterprises will still need reliable data, authorization controls, and operational monitoring.

The Linux Foundation has announced its intent to launch the Agent Name Service, or ANS, an open standard for establishing portable identities for AI agents. Built on the Domain Name System, ANS is designed to help organizations discover an agent, verify whom it represents, inspect its permissions, and determine whether its code and operational history remain authentic and unchanged.

This standard addresses a fast-growing enterprise concern as agents increasingly interact with applications, business data, customers, and other agents across organizational boundaries. Conventional application credentials identify workloads inside a single environment, but they lack a portable identity that external businesses can recognize and evaluate.

ANS takes a federated approach rather than creating a separate lookup network. Organizations could associate agents with domain names they already control, extending an established layer of internet infrastructure that processes more than 100 million queries per second worldwide. The model is intended to avoid a central gatekeeper or proprietary namespace.

“By building on DNS and open standards, ANS creates a scalable and interoperable framework for verified agent communication across the global digital economy,” said Jim Zemlin, CEO of Linux Foundation.

However, discovering an agent is distinct from trusting its automated actions. While ANS provides identity and verification signals, enterprise systems must still determine whether an agent is authorized to read a customer record, execute a transaction, alter infrastructure, or delegate work. Identity establishes who makes the request, while authorization, policy enforcement, data governance, and operational monitoring dictate the permitted actions.

The framework’s support for decentralized identifiers and Legal Entity Identifiers could make that distinction easier to manage. DIDs can connect agents with decentralized identity systems, while LEIs can tie activity to recognized legal entities. That combination may help companies evaluate both the technical identity of an agent and the organization accountable for it.

The timing reflects a broader shift in AI spending. Gartner’s April 2026 research found that organizations with successful AI initiatives invest up to 4x more, as a percentage of revenue, in foundations such as data quality, governance, and change management. Those organizations achieve up to 65% greater business outcomes in revenue growth and cost optimization.

Identity belongs in that foundation, alongside trustworthy information. Forrester’s 2026 research places data quality platforms at the forefront of enterprise efforts to scale generative and agentic AI. Separately, Anomalo argues that the data foundation comes before the AI win, emphasizing reliability and governance as organizations move beyond early demonstrations.

An agent with a verified name can still make poor decisions if its underlying data is stale, incomplete, or stripped of business context. Conversely, well-governed data alone cannot prove whether an external agent is authentic. Enterprises require both layers to protect the information supply and establish authorized usage.

Broad industry participation will determine whether ANS develops into shared infrastructure or remains one identity option among several. Representatives from GoDaddy, Cloudflare, DistributedApps.ai, Hashgraph Online, Infoblox, Salesforce, Inc., Cisco, and OWASP have expressed support for the project’s open and decentralized direction. The Linux Foundation is also seeking contributions from enterprises, AI developers, infrastructure providers, and security researchers through the Agent Name Service GitHub organization.

Practical implementation questions remain. Details surrounding how consistently agent permissions are represented across vendors, how quickly compromised identities can be revoked, and what operational evidence is sufficient to demonstrate an agent’s history remains intact will shape enterprise confidence far more than naming protocols alone.

Still, utilizing DNS gives ANS a credible starting point. The infrastructure is globally distributed, familiar to operators, and connected to domains organizations already control. If contributors can translate that reach into clear verification, revocation, and governance practices, The Linux Foundation’s initiative could help the agentic web develop around portable trust rather than siloed environments.