The co-creator of internet protocols is working on an identity standard for AI agents

The idea is that every AI agent venturing onto the open web — to search for information, order services, interact with third-party sites — would carry a verifiable "identity," much like a user or device does today. Right now most websites and services can't tell whether they're dealing with a human or an autonomous agent, and that's already causing problems, from blocking to fraud disguised as agent activity. For businesses in Kazakhstan using or planning to use AI agents for external tasks — monitoring competitor pricing, cross-checking data on supplier websites, working with external portals — it's worth building into your processes the likelihood that an agent may eventually need to officially verify it's acting on the company's behalf and in its interest. This is still a working initiative rather than a finished standard, but efforts like this typically run one to two years ahead of concrete requirements — for instance, around bot identification in e-commerce or banking APIs. Companies whose agents work purely inside their own systems (processing mail, documents, tasks in 1C/CRM) aren't directly affected, but those whose agents reach out onto the open internet should keep an eye on this.