Internet pioneer Vint Cerf is supporting an effort to create a shared identification system for artificial intelligence agents, as companies prepare for a future in which autonomous software programs interact across the open web.
Cerf, who helped design the foundational protocols of the internet, has joined Innovation Labs as an adviser after completing a 20 year career at Google. The organization is developing an open framework that could allow AI agents to verify their identities, establish their authority and operate more safely across different online services.
The initiative addresses a growing concern within the technology industry. AI agents are becoming increasingly capable of performing tasks, accessing digital resources and communicating with other systems. However, there is currently no universally accepted way to determine who created an agent, what permissions it has or who should be held responsible for its actions.
Innovation Labs is working on a proposed system called DNSid. The framework would connect an AI agent’s identity to an existing internet domain name.
Under the proposal, organizations could register their agents through domain infrastructure and use cryptographic verification to create a record of their identities. This could help websites, companies and other agents confirm where an AI system came from before allowing it to access information or complete transactions.
The approach builds on the Domain Name System, which already plays a central role in directing users to websites and online services. By connecting agents to verified domains, DNSid could provide a familiar and widely distributed foundation for identity management.
Innovation Labs is a subsidiary of Identity Digital, a company that operates domain registries. The organization is reportedly testing its proposed standards with major cloud computing and digital identity businesses, although the participating companies have not been publicly identified.
The rise of autonomous AI agents creates new questions about responsibility on the internet.
Traditional websites and domains are largely passive. AI agents, by comparison, can make decisions, initiate requests, communicate with other software and take actions on behalf of individuals or organizations.
A reliable identification standard could help clarify what an agent is permitted to do and which company or person authorized its activity. It could also make it easier to investigate harmful behavior, revoke permissions or prevent unverified agents from accessing sensitive systems.
Registration alone, however, may not solve every problem. Organizations will still need to define what guarantees they are providing when they register an agent. An identity system could confirm the source of an agent without necessarily proving that the agent is secure, accurate or trustworthy.
Many AI agents currently operate inside closed technology platforms. They generally use tools, data and services controlled by the company that developed them.
Businesses are now preparing for agents that can move across the wider internet, work with external services and communicate directly with agents created by other companies. That vision will require systems developed by competing providers to recognize and trust one another.
Without shared standards, the internet could become divided into separate ecosystems where agents built by one company cannot interact effectively with agents built by another.
Cerf believes broad adoption will depend on whether an open protocol provides clear practical benefits. The early internet developed through similar pressure, as users and organizations adopted TCP/IP because it allowed different computer networks to communicate through a common system.
Innovation Labs is positioning DNSid as neutral infrastructure rather than a proprietary platform. The organization says it does not intend to build a broader AI business around the standard or take ownership of agent registration data.
That independence could become important as companies evaluate whether to trust identity systems created by large technology providers that also compete in the AI market.
An internet filled with autonomous agents is not guaranteed, but companies are investing heavily in the possibility.
AI developers envision systems that can arrange travel, purchase products, manage business processes, negotiate with service providers and complete complex online tasks with limited human supervision. As these capabilities expand, users may increasingly delegate routine digital activities to software.
For that model to succeed, agents will need dependable ways to identify themselves and demonstrate that they are acting with legitimate authority.
The involvement of Cerf gives the DNSid initiative support from one of the most influential figures in internet history. It also highlights how the rapid development of AI is creating infrastructure questions similar to those faced during the early growth of the web.
The next stage of the internet may depend not only on making AI agents more capable, but also on ensuring that people and organizations can identify, audit and hold them accountable.
Share your thoughts about this article.
Be the first to post a comment!
