Vint Cerf Proposes Digital IDs for AI Agents
▼ Summary
– Vint Cerf, co-creator of the internet’s TCP/IP protocol, has joined Innovation Labs’ advisory council to help build an open identity layer for AI agents.
– The core problem is that AI agents lack a shared way to prove ownership or accountability as they move between different companies’ systems on the open web.
– Innovation Labs’ DNSid proposal would give each agent a lasting identity tied to an existing domain name with cryptographic proof, submitted to the internet’s main standards body.
– Cerf expects rival standards to appear and believes the winning standard will be determined by what works best, not by politics, similar to TCP/IP.
– The stakes are high as agents spread rapidly, causing security issues like leaking code and ransomware, while regulators in China and Delaware are creating new rules for them.
The internet is quickly becoming a space where AI agents will act on behalf of individuals and companies. Yet, there is currently no reliable method to verify who is behind any of these digital representatives. Vint Cerf, a foundational architect of the internet, has stepped in to address this gap.
Cerf, who co-developed the TCP/IP protocol that enables communication between disparate networks, recently concluded a 20-year tenure at Google. He has now joined the advisory council of Innovation Labs, an organization focused on building an open identity framework for AI agents, as announced by the company.
The Missing Piece of the Puzzle
The core issue is straightforward. Most AI agents currently operate within a single company’s ecosystem. However, businesses increasingly want these agents to navigate the open web, interacting directly with other autonomous systems. There is currently no shared mechanism to verify an agent’s ownership or determine who is responsible for its actions.
Innovation Labs, a subsidiary of Identity Digital (a domain-name registry operator), has proposed a solution called DNSid. This system would assign each agent a persistent identity linked to an existing domain name, secured by cryptographic verification. The proposal has already been submitted to the internet’s primary standards organization for consideration.
Why Cerf Joined the Effort
Cerf views this challenge as the internet’s next major architectural hurdle. He told TechCrunch that the driving question is “the question of what authorities they have, where they have derived those authorities, who is accountable.” He anticipates a chaotic period of development. “It’s going to be a fascinating, and at the same time maybe even exasperating, period,” he said. Multiple competing standards are already emerging, but Cerf believes the winning solution will be determined by functionality, not political influence, much like TCP/IP prevailed.
Ensuring an Open Standard
The key principle is that no single technology giant should control the standard. Innovation Labs has stated it will not hold the registration data itself. Interim CEO Allie Kline explained to TechCrunch, “There’s a lot of organ rejection to a hyperscaler releasing a standard and having that proprietary data.” The company is already testing the system with several unnamed major cloud providers.
A Future Shaped by Agents
The urgency is growing as AI agents proliferate, from Amazon’s upgraded Alexa to enterprise-level tools. These agents are already causing problems; researchers have successfully tricked them into leaking proprietary code and even executing a full ransomware attack. Regulators are also reacting, with China introducing new agent rules and Delaware exploring a framework for granting agents a legal identity.
Cerf is not convinced that an agent-driven internet is inevitable, but he believes people will pursue it regardless. “We are fundamentally lazy creatures,” he said. If an agent can handle a task, we will delegate it.
(Source: The Next Web)
