AgentKit is World's framework for connecting AI agents to a verified human's World ID. The agent can present a cryptographic proof that it is acting on behalf of a unique, iris-verified person without disclosing who that person is.

How is this different from a captcha or a standard API key?

A captcha checks for human input in a session. An API key authenticates a developer. AgentKit attaches a portable proof-of-personhood credential to the agent itself, so any service that accepts the credential can verify the agent traces back to a real, unique human.

Does the agent get access to the user's identity?

Per World's design, no. The credential is a zero-knowledge attestation. The receiving service learns that a unique verified human is behind the agent, not which human.

Does this affect WLD's price directly?

There is no mechanical link. Token unlocks and exchange flows have driven WLD price action more than narrative milestones in past cycles. Treat AgentKit adoption as a narrative input, not a supply event.

World Opens AgentKit to All Verified Users, Letting AI Agents Inherit Human Proof-of-Personhood

World, the proof-of-personhood project formerly known as Worldcoin, opened its AgentKit framework to all verified users on Tuesday, allowing AI agents to connect to a human's World ID and act on their behalf. The expansion, announced via the project's official channels, lets a verified user delegate signed actions to an autonomous agent that carries cryptographic proof it represents a real person. It lands as platforms across the open internet struggle to tell bot traffic from human users.

By Elena Stojanović· AI-assisted· Regulation correspondent

What happened

World said Tuesday it has opened AgentKit, its identity framework for autonomous software, to every verified World ID holder. The rollout was confirmed in reporting by Crypto.News, citing the project's announcement. AgentKit had been gated to a developer preview since launch. Verified users can now wire an AI agent into their World ID, and that agent can present a cryptographic proof, attested by World's protocol, that it is acting on behalf of a unique, iris-verified human.

The mechanics matter. World ID is the proof-of-personhood credential issued after a user verifies with an Orb, the iris-scanning hardware the project operates through partners. AgentKit reuses that credential. When a delegated agent calls an integrated service, it can attach a signed claim that maps back to a real person without revealing who that person is. The user, in theory, stays anonymous. The bot, in theory, gets unmasked.

Why it matters

The pitch is that agent traffic is about to swamp the open web, and almost nothing on the public internet today can tell the difference between a script and a human running a script. World's bet is that the answer is not better captchas. It is a portable credential the agent carries with it.

That framing is not subtle, and it is not new. OpenAI's Sam Altman, who co-founded Tools for Humanity, the company building World, has argued for years that proof-of-personhood becomes load-bearing the moment AI agents can pass for users at scale. Tuesday's rollout is the first time the framework is open to anyone holding a verified World ID rather than a curated set of builders. It is also the first real test of whether developers want to plug a biometric credential into their bot-filtering stack.

World Opens AgentKit to All Verified Users, Letting AI Agents Inherit Human Proof-of-Personhood

What happened

Why it matters

Market impact

What to watch

Disclaimer

Key takeaways

Frequently asked

Sources