Sam Altman's World Launches AgentKit – Verifying Humans Behind AI Shopping Agents

A New Standard for Agentic Commerce

Tools for Humanity (TFH), the startup behind World – Sam Altman's biometric identity project – has launched a beta of a new tool called AgentKit. The goal is to solve a rapidly growing problem: Who is actually responsible when an AI agent shops on your behalf?

The Problem: Who's Really Buying?

"Agentic commerce" – the practice of using AI programs to browse the web and make purchases on a user's behalf – is growing explosively. But it raises serious questions about accountability, fraud, and privacy. When an AI agent places an order, is it:

• A user who has explicitly approved it?
• A runaway bot?
• An attacker who has hijacked the agent?

AgentKit – The Solution

World's AgentKit uses the company's existing biometric infrastructure (the iconic iris-scanning orb) to verify that a real human has actually authorized the transaction. This creates a cryptographic chain from human to agent to action.

For retailers and payment platforms, this means:

• Lower fraud risk for agent-based purchases
• Clearer accountability in dispute resolution
• Compliance with consumer protection regulations

CIO Relevance

For organizations building agentic workflows, this is directly relevant. As AI agents gain more autonomy – from ordering office supplies to approving invoices – identity verification will become critical infrastructure.

TFH is positioning World not just as a crypto/privacy project, but as the foundation of the agentic internet. It's an ambitious but logical extension of biometric identity into a world where AI acts on behalf of humans.

AgentKit is currently in beta, available in English in the US.