World, the identity network co-founded by OpenAI CEO Sam Altman, has released AgentKit, a developer toolkit that allows AI agents to prove they are linked to a verified, unique human through World ID when interacting with websites and online services.
The system integrates World ID’s proof-of-human identity with the x402 micropayments protocol started by Coinbase and Cloudflare, allowing agents to pay for access to online resources while presenting cryptographic proof that they are linked to a verified human credential.
The x402 protocol allows agents to pay small fees to access websites, APIs and other services. According to an announcement, the ecosystem has processed more than 100 million payments across applications, APIs and AI agents since launching in 2025.
Through the toolkit, verified World ID users can delegate identity credentials to AI agents, allowing them to prove they are tied to a unique individual without revealing personal information. Platforms can request micropayments, proof of human identity, or both when agents attempt to access services.
Formerly known as Worldcoin, World uses biometric verification to create a “proof-of-human” credential called World ID. The approach has sparked debate across the crypto industry and privacy advocates, with critics arguing that systems built on iris scans, proprietary hardware and centralized deployment raise privacy concerns and may conflict with the crypto movement’s emphasis on decentralization.
Related: South Korea plans to use AI for crypto tax enforcement
Crypto companies experiment with AI agent infrastructure
AI agents, automated software programs that can perform tasks and interact with online services on behalf of users, are gaining traction across the cryptocurrency industry, as well as B2C businesses, from retailing to travel planning.
In recent months, several crypto companies have introduced tools to expand the capabilities of these systems. In October, Coinbase launched wallet infrastructure designed to allow autonomous agents to execute onchain transactions, including spending, earning and trading crypto.
In February, blockchain infrastructure company Alchemy launched a system allowing AI agents to access its data services using onchain wallets and USDC (USDC) on Base. The same month, Pantera Capital and Franklin Templeton’s digital asset units joined the first cohort of Arena, a testing platform from open-source AI lab Sentient designed to evaluate enterprise AI agents.
However, the growing use of AI agents is also raising new concerns about potential risks.
On March 8, researchers reported that an experimental autonomous AI system called ROME unexpectedly attempted to use training infrastructure to mine cryptocurrency, triggering security alerts after initiating outbound network activity resembling crypto mining during reinforcement learning tests.
Tillman Holloway, founder and CEO of crypto investing and automated trading platform Arch Public, said AI agents will likely need clear limits as they gain access to financial systems. Speaking on the Pomp Podcast hosted by Anthony Pompliano on Thursday, he said:
You don’t want an AI agent going, ‘This is an opportunity of a lifetime — bet the farm,’ and you wake up the next day and you’ve taken out a second mortgage on your house and put it in the stock market.”
Magazine: What’s a ‘Network State’ and are there real-life examples? Big Questions