Your AI agent can book a flight, pay a supplier and message a client in your name. So how does the other side know it's genuinely yours, and not an impostor wearing your agent's name? On November 10, 2026, the Linux Foundation welcomed a project built around that exact problem: the Agent Name Service, or ANS.

ANS is an open standard that gives AI agents a verifiable identity, along with discovery and authentication, and it does it using infrastructure the internet already trusts. Think of how websites earned credibility. DNS gave them names machines could resolve, and certificates let a browser confirm a site is who it claims to be. ANS anchors an agent's name to a domain in the same spirit, so the identity traces back to an organisation that controls that domain.

The mechanics cover domain-anchored naming, registration, DNS publication, public-log auditing, integrity monitoring and agent-to-agent authentication. In practice, that means an agent's record can be looked up before anyone trusts it, changes to that record leave a public trail, and two agents meeting for the first time can check each other before money or data changes hands.

Why now? Agents are being handed real authority — spending money, signing off on orders, speaking to customers — while identity checks lag behind. A verified name doesn't make an agent smart, but it makes impersonation far harder, and impersonation is exactly the trick attackers reach for first.

The support list is hard to ignore. The project launched with seven specification documents and 11 supporting organisations, including Cisco, Cloudflare, DigiCert, GoDaddy, Salesforce and OWASP. Networking, DNS, certificates and security are precisely the skills an identity system like this needs. The announcement went out through PR Newswire on November 10, and Unite.AI and ITBrief covered it the same week.

One honest caveat before anyone gets carried away: the specifications are still labelled drafts. Details can shift, implementations are early, and the standard only becomes real if agent platforms and businesses actually adopt it.

I wrote up the full launch, including why identity is fast becoming the missing layer in agent security, here: Linux Foundation Launches Agent Name Service to Give AI Agents a Verifiable Identity