World links iris scanning and proof-of-humanity to AI agents meant to transact
World, the company co-founded by Sam Altman that scans people's irises to confirm they are real, announced new tools this week that let people link their "proof of humanity" to an AI agent.
World, the company co-founded by Sam Altman that scans people's irises to confirm they are real, announced new tools this week that let people link their "proof of humanity" to an AI agent. The goal is agentic commerce: instead of platforms blocking all robots, a verified person could vouch for their agent. The background is concrete — several large US banks are already restricting third-party agents because they cannot tell which ones a customer actually authorized, according to Axios reporting cited by Semafor.
What happened
This week World announced tools that let users attach their proof of humanity to an AI agent, reports J.D. Capelouto in Semafor. The company's chief product officer Tiago Sada pointed to what this could mean for commerce: in the future, verified agents could give sellers reason to offer perks like discounts and free trial periods. "If you know that an agent is coming from a human, you can actually be much more generous," Sada told Semafor.
This is Sada's and World's expectation of a possible future, not an established practice. No sellers are yet documented offering such perks. But the news lands in a week in which World has rolled out its entire verification stack into commercial channels at once.
The week that shows the strategy
On October 19, World ID for Zoom launched in general availability. Verified participants in video calls get a "Human Badge" next to their video feed. Setup requires a physical iris scan at an Orb device plus a verification selfie. World ID for Zoom gives enterprise customers access to a verification process that can distinguish real video from AI-generated content, according to TechRadar.
On the finance side, World names, according to Decrypt, Bain Capital, OpenExchange, Outtake, Pantera Capital and VanEck Funds as the first design partners for World ID in banking and finance.
The timing fits the market World is pointing toward. Gartner research cited by TechRadar shows that 36% of senior cybersecurity leaders reported at least one deepfake incident in a video call over the past twelve months. World also highlights Deloitte's forecast that AI-enabled fraud in the US could reach $40 billion in 2027, up from $12.3 billion in 2023 — figures the company itself uses in its marketing.
How the stack works
World offers three cryptographic proofs, described by Decrypt:
- Proof of Human: confirms that someone is a real and unique human through a single visit to one of World's Orb devices, which scan the iris.
- Proof of Presence: a selfie-based liveness check on the phone — a simpler verification that you are alive here and now.
- Proof of Attributes: confirms details like nationality or age from a government ID document, revealing only the requested information.
The key mechanism for agent use is AgentKit, a developer toolkit still in beta, which lets users delegate specific proofs to AI agents so they can transact on a person's behalf. It is this delegation logic that makes "vouching" possible: the agent is not a human, but it carries a cryptographic proof that a verified human stands behind it.
One open question remains: the reporting does not specify exactly which "new tools" were announced this week, or whether they are identical to the AgentKit beta from October 19. It could be the same toolkit, or a further development — the sources do not resolve this.
Why banks are blocking — and what World wants to replace the blocking with
The problem World is attacking is concrete. According to Axios reporting cited by Semafor, several large US banks are restricting interaction with third-party agents, in part because they cannot distinguish which agents the customer actually authorized. Since platforms have had no way to verify this, they have in practice had to block all robots.
World's claim is that proof-of-human technology can let a person stand accountable for their agents, so that malicious bots can be kept out without platforms having to ban all of them. For banks, watching fraud figures grow, the point is being able to answer one question: does this agent come from my customer, or from an attacker?
Scale — and who is saying it
According to World, more than 18 million people have a Proof of Human ID, and the World Foundation has raised $52.5 million this year to scale the network. Both figures are the company's own claims, relayed via Decrypt, and are not independently verified.
The regulatory aftermath
World's foundation — iris scanning against identity credentials — has met significant resistance from privacy regulators, according to Decrypt. Hong Kong's privacy regulator ordered the project to stop its iris scanning in 2024. The same year, South Korea's privacy authority fined the project roughly $830,000. In December that year, Spain's data protection authority demanded that all iris data be deleted. A Kenyan court has also ordered data deleted, and in Brazil a ban on the project was upheld on appeal — all according to Decrypt.
This history limits how quickly "verified human" can become commerce infrastructure: the technology meant to make your agent credible is itself contested in several jurisdictions.
The unresolved
Sada's vision of discounts and free trials for verified agents is, for now, a sales pitch. Whether sellers will actually build such incentives, what the pricing looks like, and how the seller side is technically integrated are not documented in the available reporting. The same applies to the relationship between the AgentKit beta and the tools announced this week. What is clear is the direction: World wants proof of humanity to become the infrastructure agentic commerce rests on — and it is racing against time, against banks' blocking practices, and against its own regulatory problems.