World, co-founded by Sam Altman, has rolled out the beta version of AgentKit, a verification tool designed to ensure that AI programs making purchases online are acting on behalf of real humans. The move addresses growing concerns about fraud, spam, and misuse in agentic commerce, where AI agents browse websites and complete transactions for users automatically.
AgentKit integrates with World’s existing verification system, World ID, which uses the company’s Orb device to scan a user’s iris and create a unique, encrypted digital code. This verified ID is then used to confirm human authority behind an AI agent’s actions, giving commercial websites a way to distinguish authentic transactions from potential abuse.
The software also works with the recently launched x402 protocol, a blockchain-based standard developed by Coinbase and Cloudflare, allowing AI programs to transact directly online. By linking a registered AI agent to a World ID, websites can verify that a human approves the purchases, adding a critical layer of trust to automated commerce.
TFH Chief Product Officer Tiago Sada likened the system to granting “power of attorney” to an AI agent, emphasizing that websites can still choose to block agents they suspect are acting in bad faith. “What the World ID badge tells you is that someone is a real and a unique human,” he said, highlighting the tool’s role in safeguarding the integrity of online transactions.
The beta release comes as major e-commerce and financial platforms, including Amazon, Mastercard, and Google, increasingly adopt agentic commerce capabilities. With AgentKit, World positions itself as a standard-bearer for secure, human-verified AI transactions, offering developers and consumers a way to embrace automated online shopping without compromising reliability or trust.