AI Agents Are Already Acting on Behalf of Companies
AI agents are no longer an experiment. They search for information, compare offers, fill in forms, and interact with other systems on behalf of companies. According to WEF (World Economic Forum, an international organisation for economic cooperation) data, 82% of executives plan to adopt AI agents within the next one to three years.
As a result, a question arises that the internet has never had to answer at this scale. When an AI agent performs an action, the agent itself is not liable for anything. The legal entity behind it is. But how can anyone verify which legal entity that is? Today’s internet has no common standard for it.
What Is the Agent Name Service?
In June 2026, the Linux Foundation announced its intent to launch ANS (Agent Name Service, an open standard for verifying the identity of AI agents). According to the Linux Foundation press release, the goal is to create a trusted layer for identifying, verifying, and discovering AI agents across the internet.
ANS does not build new infrastructure from scratch. Instead, it relies on DNS (Domain Name System, the internet’s naming system that translates domain names into addresses), which has served as the foundation of the internet for over forty years. In practice, this anchors agent identity to domain names that organisations already own. There is no central registry and no system that a single company controls.
The ANS framework makes three things verifiable. First, who the agent represents. Second, what permissions it has. And third, whether its code and operational history are authentic and unchanged.
However, an important clarification: ANS is currently an intent, not a live service. The community develops the standard in the open and publishes the technical repositories on GitHub. Cloudflare, Cisco, Salesforce, GoDaddy, and Infoblox, among others, support the initiative.
Where Does the LEI Fit In?
ANS works with existing identity systems by design. The press release names two of them: the DID (Decentralized Identifier, a decentralised digital identifier) and the LEI (Legal Entity Identifier, the global identification code for legal entities).
That detail matters. Technical identity answers the question of which agent this is. Legal identity, in turn, answers the question of who is responsible for it. These are two different layers, and the agentic web needs both.
The LEI already does the work of identifying legal entities today. It is the only global standard that covers all legal entities regardless of country or legal form. The LEI code answers who is who and who owns whom, and anyone can verify the data in the GLEIF register. Therefore, when an AI agent references an LEI in its identity, any party can check which company stands behind the agent.
We have written about the broader connection between the LEI and artificial intelligence in our earlier article LEI code and AI. ANS is the next concrete step in that development: for the first time, an internet infrastructure standard includes the LEI directly, not only financial regulation.
What Does This Mean for Companies?
For now, nothing mandatory. ANS is in the planning stage, and it creates no new obligations for companies. In addition, the LEI is not the only or a mandatory identifier in ANS. It is one of the supported layers.
Still, the direction is clear. Verified organisational identity is moving beyond financial regulation into the foundations of the internet itself. The same pattern repeats elsewhere: for example, the role of the LEI is growing in payment standards, in business verification, and in digital identity initiatives in Europe and beyond.
For a company that wants to be ready, the first step is simple: a valid LEI code. If you do not have one yet, you can register an LEI in a few minutes. If you already have one, make sure it is renewed, because no system can rely on expired LEI data.
What Happens Next?
The development of ANS is only beginning. The technical content of the standard, its timeline, and its adoption will become clear over the coming months and years. Meanwhile, participation is open to companies, developers, and infrastructure providers.
We are following developments at both the Linux Foundation and GLEIF and will cover the most important steps ahead. Because the question of which legal entity is behind an AI agent is not going away. And the answer starts with a valid LEI code.