World, the identity project co-founded by Sam Altman, launched a World ID suite for banks and financial institutions on Sunday, October 18, 2026, at the Money20/20 conference, according to The Block. It is World's first product aimed at a single industry. The pitch is simple: deepfakes and AI impersonation are making it hard for a bank to know whether the person on a video call, or the software agent calling its API, is who it claims to be, and World wants to be the "proof of human" layer.
What is confirmed is narrower than the headline. The suite bundles three existing products, one of them (the Zoom check) is now generally available, and no bank has been named as a customer. This post separates the shipped pieces from the claims, explains how the technology is supposed to work, and lists what a risk team would need to see before relying on it. For the wider picture of AI-driven fraud, see our look at the $25 million deepfake video-call scam and our claim check on the AI scam economy figures.
TL;DR: what is claimed and what is verified
| Question | Answer | Status |
|---|---|---|
| What launched? | A World ID suite for banks and financial firms, unveiled by co-founder Alex Blania | Reported by The Block and others |
| When? | Sunday, October 18, 2026, at Money20/20 (some outlets date it October 19) | The Block dates it October 18 |
| What is in it? | World ID app, World ID for Zoom, World ID for Agents (AgentKit) | Components already existed |
| Is anything new and generally available? | World ID for Zoom is now GA after an April limited beta | Reported by The Block |
| Which banks use it? | None named | Not disclosed |
| Price? | Not disclosed | Not disclosed |
| Design partners | OpenExchange, Bain Capital, VanEck Funds, Outtake | Named by World, to "explore" uses |
| Does it work against deepfakes? | World says so; no independent test found | Unverified |
| Scale | More than 18 million Proof of Human holders | World's own figure |
What exactly is in the suite?
The three parts, as described by The Block and a CoinCodex write-up of the launch:
1. The World ID app. An identity wallet available in 150-plus countries that holds cryptographic "Proofs." CoinCodex lists three: Proof of Human (a unique, real person, which requires a one-time visit to World's Orb scanning device), Proof of Presence (a phone-based liveness check), and Proof of Attributes (a specific government-ID fact such as age or nationality, shared without revealing other data).
2. World ID for Zoom. A check that shows a "Human Badge" on the video tile of a verified participant, so you can tell a real person from a deepfake. It began as a limited beta in April and is now generally available. Tools for Humanity, the main developer behind World, also joined Zoom's independent software vendor exchange program, per The Block.
3. World ID for Agents (AgentKit). Lets a person delegate Proofs to an AI agent, which can then show a merchant or bank that it acts for a verified human without revealing who that human is. It is still labelled beta. In March, World said it had released AgentKit in beta using Coinbase's x402 protocol, per The Block.
Identity badge on a lanyard representing proof of human verification for banks
Why banks, and why now?
World frames the product around impersonation. The Block cites a Deloitte forecast that US AI-enabled fraud losses could reach $40 billion by 2027, up from an estimated $12.3 billion in 2023. That number comes from Deloitte and describes the problem rather than the product, but it explains the sales pitch. A bank's biggest exposures are exactly the places the suite targets: account recovery, large wire approvals, video-based onboarding and executive impersonation on calls.
The agent angle is the newer idea. As software agents begin to move money and call APIs, banks have to decide how to treat automated traffic that is legitimate. World argues a Proof of Human attached to an agent could let a bank say "this agent acts for a verified person." CoinCodex notes World says many sites currently block automated traffic, which is the gap it wants to fill. Treat that as World's argument, not an industry standard.
This follows a pattern we have tracked: real-time impersonation is getting cheap, which is why we covered Tavus's video Turing test, and why media-provenance tools such as Google's SynthID verification site and C2PA Content Credentials are getting attention. Those tools check a piece of media. World instead tries to check the person.
How the privacy claim is supposed to work
According to CoinCodex, World says institutions receive zero-knowledge proofs instead of the underlying personal data. A bank could confirm nationality, or require a verified human to approve a large wire, without collecting or storing ID documents. That is the design claim; we did not find an independent audit of the bank-facing flow, and CoinCodex's article cites no outside expert or critic.
A zero-knowledge proof lets someone show a statement is true (this account holder is a unique human over 18) without handing over the underlying data. The cryptography is well studied. The practical risks lie elsewhere: how the Orb enrollment is secured, who can revoke a credential, what happens when a phone is lost, and what a bank logs on its side. Those are the questions a security review would ask.
A peeled mask revealing a seam, illustrating deepfake detection in a banking video call
Claimed versus verified
Verified by reporting we could read:
- The suite launched on October 18 at Money20/20, unveiled by Alex Blania (The Block, CoinCodex).
- World ID for Zoom is generally available; the beta ran from April (The Block).
- The named design partners are OpenExchange, Bain Capital, VanEck Funds and Outtake (The Block). VanEck was the only finance firm World named among its April Zoom beta testers, per The Block's reading of a World blog post.
- Tools for Humanity president Sandro Herbig described the offering as a set of tools "to add an AI-safe trust layer" across internal and external systems (The Block).
World's claims, not independently checked:
- More than 18 million people hold a Proof of Human. CoinCodex also repeats a World figure of 475 million presentations.
- That the Zoom badge reliably separates humans from deepfakes in a live call.
- That zero-knowledge proofs give banks what they need without storing documents.
Not disclosed:
- Any bank currently using the product.
- Pricing.
- What the design partners are actually building or testing.
- Whether regulators will accept Proof of Human for any compliance purpose. Coverage says it is meant to sit alongside KYC, not replace it.
We did not find a primary announcement post on World's own site in our searches, so the launch details here rest on The Block and other outlets that relay World's statements. The Block said it had asked World for comment and had not received a response when it published.
What a bank (or any company) should ask before adopting
- What is the threat model? A badge on a Zoom tile stops a deepfake only if the attacker cannot also present a valid badge. Ask how credentials are bound to a person and how stolen or coerced enrollment is handled.
- Who is the root of trust? Proof of Human depends on a single biometric enrollment network. A bank would be tying its identity checks to one vendor's hardware and policies.
- Fallbacks and exclusions. Users who cannot or will not visit an Orb still need service. What is the fallback, and does it become the weak point attackers target?
- Regulatory fit. Biometric data rules differ across jurisdictions. Our overview of AI regulation in the EU and US is a starting point.
- Evidence. Ask for third-party testing against current deepfake tools, with false accept and false reject rates, before any pilot.
How to protect yourself now
Whatever vendor a bank picks, the basics for individuals and small teams do not change. Call back on a known number before acting on any urgent money request, even on video. Agree a code word for finance approvals. Treat a "verified" badge as one signal, not proof. And if your company lets AI agents act on accounts or spend money, put guardrails in front of them: AgentBeam is the agent security platform from the explainx.ai team that stops AI agents before they take dangerous actions.
What people are asking
Is World ID the same as Worldcoin?
World is the project formerly known as Worldcoin, and Tools for Humanity is the main company developing it. This post covers the identity product only, not the project's token.
Why does Sam Altman's name come up?
He is a co-founder of World. The launch coverage cites that link, but the banking suite is a World and Tools for Humanity product, not an OpenAI one.
Is this the same as a government digital ID?
No. Governments are building their own: see our post on China's digital human ID card pilot. World ID is a private credential with no government issuer, which is both its pitch (no state dependency) and its open question (no legal status).
Does this connect to AI-enabled cyberattacks more broadly?
Yes in spirit. Bill Gates recently warned about AI cyberattacks as a leading risk; we summarised that in Gates on AI cyberattacks. Identity is one of the attack surfaces AI makes cheaper.
What we are watching
- A named bank or payment network piloting the suite, and what it says publicly.
- An independent evaluation of the Zoom badge against current real-time deepfake tools.
- Pricing and the enterprise terms.
- Any regulator statement on whether a Proof of Human counts toward customer due diligence.
- A primary announcement from World with technical documentation for the bank-facing flow, which we will add as an update when it appears.
Related reading
- The $25 million deepfake video-call scam and how to verify
- AI scam economy numbers: a claim check
- Tavus and the video Turing test
- Google SynthID verification site guide
- What is C2PA Content Credentials?
- China's digital human ID cards: claimed vs verified
- Bill Gates on AI cyberattacks
Details are as reported on October 18 and 19, 2026 and may change as World publishes more.