Agent Anarchy: Why AI Agent Identity Is Becoming a Problem

Imagine you're an e-commerce owner and your server starts getting flooded with thousands of requests from an unknown AI agent. Who is it? Who's behind it? How do you stop it? Today, there are no answers to these questions — and that needs to change before the agent economy kicks into full gear. AI agents are evolving from research experiments into real economic actors. On July 15, DoorDash launched a limited beta of dd-cli, a command-line tool that enables AI agents to browse restaurants, assemble orders, and purchase food. Adobe, meanwhile, announced it is integrating agent commerce capabilities into its Adobe Commerce platform — used by brands like Coca-Cola and Major League Baseball. "Organizations need a common way to recognize who is responsible for an AI agent, regardless of where the agent operates," says Allie Kline, interim CEO of Innovation Labs, the company behind the project. But the problem runs deeper than one or two agents. Tens of millions of agents like OpenClaw or NanoClaw freely roam the web today with a single form of identification: an IP address. In an era where bad actors are plentiful, that's not enough.

Vint Cerf Brings the Weight of Internet History

The DNSid project has now been joined by Vint Cerf — a legend of internet infrastructure who co-created the TCP/IP protocols in the 1970s, on which the entire internet still runs today. This isn't the first time he's been involved in solving fundamental architectural challenges of the network, but this time it's about the future, not the past. "AI agents are driving the next major architectural evolution. Questions of identity, accountability, and interoperability require the same thoughtful architectural approach that enabled the internet to scale," Cerf said. His involvement lends the project considerable credibility — few people understand scaling global systems better than the person who helped build them from the ground up. The DNSid advisory board also includes leaders from the fields of internet infrastructure, cybersecurity, national security, and finance. This signals that the project is not just an academic exercise, but an ambition to lay the groundwork for something that could become an industry-wide standard.

DNS as the Foundation of Identity: Good Old Infrastructure in a New Role

The key idea behind DNSid is elegant simplicity: don't build anything new, but leverage what already works on a global scale. The Domain Name System (DNS) — a decades-old technology that translates domain names into IP addresses — is to serve as the anchor for the permanent identity of AI agents. Why DNS? Because it's one of the few systems that already reliably works across organizational boundaries. Independent organizations can recognize the same identifier without needing a direct relationship with each other. That is exactly the type of foundation that AI accountability will need. In June, Innovation Labs submitted a standards proposal to the IETF (Internet Engineering Task Force), which has been managing core internet protocols for decades. The approach is pragmatic: instead of inventing a new trust system from scratch, DNSid builds accountability on infrastructure that already works, and plans to further develop it through open standards.

A License Plate, Not a Criminal Record

One of the most important distinctions DNSid emphasizes: identity is not the same as trust. The project does not aim to decide which agents are trustworthy or authorized to act. Its goal is to determine who is responsible. What companies, security teams, regulators, and courts do with that information is up to them. In other words: DNSid is a license plate, not a criminal record. It's a foundational layer, not a gatekeeper. This philosophical approach reflects proven practice from internet history — first build neutral infrastructure, on which more sophisticated decision-making layers can later be constructed. "DNSid does not decide whether an agent or its owner is trustworthy. It determines who is responsible," Kline explains. "That gives counterparties the information they need to make their own trust decisions."

What DNSid Means for the European and Czech Landscape

For Czech companies and the European market, this initiative holds particular significance. The EU AI Act, which entered into force in 2024 and whose requirements are being phased in gradually, emphasizes transparency, traceability, and accountability of AI systems. A project like DNSid could provide the technical infrastructure that helps companies meet regulatory requirements in practice — not just on paper. For Czech developers and startups working with or building AI agents, it's crucial that DNSid is intended to be an open standard, not a proprietary technology. This means that no single company — whether OpenAI, Google, or Anthropic — would define how accountability should work in the era of agentic AI. "The best standards fade into the background. Developers shouldn't have to build a different accountability integration for every AI platform. A common standard will let them build once, work everywhere, and focus on building great AI instead of rebuilding the same infrastructure over and over," Kline adds.

Open Questions and the Path to Adoption

The biggest question mark remains industry adoption. The largest AI labs — OpenAI, Anthropic, Google DeepMind, Meta — will be crucial for any real-world solution. Innovation Labs, however, champions the philosophy that the open standards model that enabled the internet to scale is the right template for scaling AI as well. Of course, even the best standard won't prevent malicious actors from voluntarily avoiding identification. But the value of a widely adopted standard lies in enabling systems to recognize the absence of identity. Over time, the inability to verify who stands behind an AI agent should become a clear warning signal — much like websites without HTTPS certificates today.

What the Future Holds

The DNSid project arrives at the right time. Agentic AI is rapidly moving from demos to production deployment — from ordering food, to managing finances, to critical infrastructure. Without an accountability layer, we risk a world where autonomous systems act in a vacuum with no way to trace responsibility. Vint Cerf's involvement gives the project not only weight but also historical perspective. The internet as we know it didn't emerge as a perfect plan, but as a set of gradually built open protocols. DNSid seeks to follow the same path — and if it succeeds, it could become one of the pillars on which a trustworthy agent economy stands for decades to come.