World Launches AgentKit to Let AI Agents Carry Proof of Human Backing

World — the crypto project co-founded by Sam Altman — announced today that it has launched AgentKit beta, a developer toolkit that extends its World ID proof-of-human system to AI agents, according to a press release shared with The Defiant.

AgentKit integrates with x402 — an payments protocol for AI agents developed by Coinbase and Cloudflare — and enables verified humans to cryptographically delegate their World ID to AI agents. The result is what World calls a “human-backed agent”: an automated actor that can prove a unique real person stands behind it, without revealing who that person is.

“Payments are the ‘how’ of agentic commerce, but identity is the ‘who,’” said Erik Reppel, Head of Engineering at Coinbase Developer Platform and Founder of x402. “By integrating World ID with the x402 protocol, developers now have a complete trust stack.”

Back in early 2023, when World ID launched alongside GPT-4, the project was already positioning itself as a counterweight to the coming wave of AI-generated accounts and automated activity online. The core pitch has barely changed — what has changed is the urgency. AI agents are now a reality, not a forecast.

The project has repeatedly attracted regulatory intervention globally over its biometric data collection practices, with suspensions and investigations in Kenya, Spain, Portugal, Hong Kong, and South Korea. As recently as last year, the High Court of Kenya ruled that Worldcoin’s collection of biometric data from Kenyan citizens in 2023 was illegal and violated the country’s data protection laws. Spain also mandated deletion of all iris scan data collected there, citing inadequate data handling practices.

Those concerns don’t disappear with AgentKit. While World frames its zero-knowledge architecture as privacy-preserving — users prove uniqueness without sharing personal information — critics have long argued that building a global identity layer on top of biometric iris scans introduces systemic risks that clever cryptography alone cannot resolve. The prospect of that same biometric infrastructure being extended to a sprawling ecosystem of autonomous AI agents is likely to sharpen that debate further.

World reports that its network now includes nearly 18 million verified humans across more than 160 countries. AgentKit beta is available now to developers who hold a verified World ID. The current release is built on existing World ID architecture, with a more advanced version planned as the next generation of the protocol rolls out.

The Rise of the Agent Economy

The announcement arrives at a pivotal moment for on-chain agentic activity. Since the start of 2026, the number of agents using the ERC-8004 standard across blockchain networks has grown from 337 to nearly 130,000 — an increase of over 39,000%, as The Defiant reported this week. As that explosion in agent activity has accelerated, so have questions about trust, accountability, and how platforms can distinguish between legitimate users and coordinated bot swarms.

The broader agentic economy continues to accelerate. Circle recently launched Nanopayments on testnet, offering gas-free USDC transactions designed specifically for AI agents making rapid, sub-cent payments for services like pay-per-call APIs and machine-to-machine marketplaces.

CoinFello last week released an open-source skill allowing agents to execute on-chain transactions via MetaMask without ever accessing a user’s private keys —addressing a core security vulnerability in how most agent wallets currently operate.

World’s AgentKit adds a third pillar to this emerging stack: alongside payments and secure key management, agents can now carry verifiable proof that a real human is behind them.

This article was written with the assistance of AI workflows. All our stories are curated, edited and fact-checked by a human.