The Man Who Created the Internet Has a New Mission: Give Identity to AI

The evolution of artificial intelligence agents may require an identification infrastructure similar to that which enabled the spread of the Internet. This is the belief of Vint Cerf, one of the designers of the TCP/IP protocols and a historical figure in the development of the web, who after twenty years at Google has decided to take on a new role as a consultant for Innovation Labs, an organization committed to defining an open architecture for AI agents' identity. The news was reported by Techcrunch.

Innovation Labs is a subsidiary of Identity Digital, a company specializing in DNS registry management, and aims to leverage the domain name infrastructure to create a system capable of identifying, authenticating, and tracking artificial intelligence agents. The idea stems from the perspective that, in the future, a growing share of online interactions may take place directly between software agents, rather than between human users.

Currently, most AI agents operate within proprietary ecosystems, relying on services and resources controlled by the same company that develops them. However, several enterprises are envisioning scenarios in which these systems could operate much more autonomously on the Internet, collaborating with agents developed by different organizations. One of the main obstacles, however, remains the lack of a shared standard to identify and verify their reliability.

To address this problem, Innovation Labs has proposed DNSid, a registry designed to associate each AI agent with an existing Internet domain. The system uses cryptographic proofs to document the registration and identity of the agent over time, providing a mechanism that should promote audits, accountability, and verification of the permissions under which agents operate.

According to Interim CEO of Innovation Labs, Allie Kline, the technology is already being tested with various hyperscalers and companies specializing in digital identity, although the names of the involved partners have not been made public.

Cerf believes that the issue of identification will become central with the spread of AI agents. "I thought I could contribute at a time when naming and identification are becoming increasingly important," he stated. "Everything arises from the emergence of artificial intelligence agents and the questions about their permissions, the provenance of those permissions, who is responsible for an agent's behavior, how its identity is established, and why one should trust it."

According to the pioneer of the modern Internet, defining a shared system will be anything but simple. Unlike traditional domain names, AI agents can perform much more dynamic and autonomous activities, making it less clear what responsibilities are assumed by the organizations that register them.

Cerf anticipates a phase of intense experimentation, describing it as "fascinating and at the same time perhaps even exasperating," precisely because of the enormous functional potential of these systems. Another challenge concerns interoperability. If every company were to adopt incompatible proprietary technologies, the risk would be to fragment the AI agents market. For Cerf, it will be user pressure that drives the convergence towards common standards, similar to what happened over the years with TCP/IP.

As for the future of the economy based on AI agents, Cerf does not speak of an "inevitable scenario," but acknowledges that businesses and users will actively attempt to delegate an increasing number of tasks to autonomous systems. "I don’t believe it is inevitable," he stated. "But I believe it is inevitable that people will try. We are fundamentally lazy creatures, and if we can find a way to have an agent do something in our place, it’s very likely we will choose it because it’s simply easier.