The rise of personal AI agents, like those built with tools such as OpenClaw, has created a new problem for the web. While a single user can deploy a small army of bots to book a table or buy tickets, the resulting traffic can overwhelm online services, mimicking disruptive denial-of-service attacks.

World, the identity startup originally known for its WorldCoin cryptocurrency project, believes it has an answer. The company has moved into beta testing for a system that ties AI agents to verified human users. The goal is to let websites distinguish between a malicious bot swarm and an authorized agent working for a real person.

The foundation is World ID, a digital identity protocol that has verified roughly 18 million people globally through physical iris-scanning devices called Orbs. With its new Agent Kit, World enables those users to cryptographically link their proven identity to any AI agent they operate.

This approach suggests a shift from blunt blocks on automation. Instead, a site could require an AI agent to present a valid World ID token. This would grant the agent permission to perform specific, limited actions—securing a reservation, entering a ticket queue, or participating in an online poll—with the confidence that a single human is accountable for the request. For platform managers, it offers a potential path to manage automated traffic without shutting it down entirely.

Source: Ars Technica