World's co-founder Alex Blania stood up at the Money20/20 conference in Las Vegas on Oct. 18, 202613. He pitched banks a defense against a costly crime. One firm had already lost $25 million to it. He called the product World ID for banking and finance. It asks one question: is the face on this call a real human? The launch marks a turn for World, too. It is the company's first product built for a single industry since it launched its identity protocol three years ago. Sandro Herbig leads Tools for Humanity, the company that builds World's technology. He framed the stakes plainly. “As AI becomes increasingly prevalent in all aspects of our lives, enterprises are requesting individuals provide 'Proofs' about themselves to help mitigate fraud and increase trust,” he said.
A Three-Way Match Built for the Call That Cost Arup $25 Million
Arup is a London-based engineering firm with offices across Asia4. On Feb. 2, 2024, a finance worker in its Hong Kong office joined a video call. The callers looked and sounded like the firm's chief financial officer and several colleagues. Every one of them was fake. Fraudsters built the deepfakes from public recordings of the real executives. The worker authorized wire transfers that totaled $25 million, or roughly HK$200 million. Discovery came only after the money was gone. Hong Kong police opened an investigation3. As of early 2025, investigators had announced zero arrests, and the money stayed gone.
World ID for Zoom answers that exact scenario with a three-way check13. It compares three images. One is the face captured when a person first scanned their iris at a World Orb. Another is a fresh selfie from their device. A third is the live video feed in the call itself. Matching all three earns the caller a "Verified Human" badge on their video tile. Any participant can request a fresh check mid-meeting too. That detail matters. Fraudsters tend to spring a deepfake once trust has settled into a call, well past its opening seconds.
The tool moved out of a limited beta that started in April 20261. It reached general availability the same week Blania spoke. Tools for Humanity also joined Zoom's independent software vendor exchange program. That move puts the check inside the Zoom Marketplace, a step up from a separate app.
Both sides, in their own words
Does a check built for the face catch a crime built on the voice?
“As AI becomes increasingly prevalent in all aspects of our lives, enterprises are requesting individuals provide 'Proofs' about themselves to help mitigate fraud and increase trust”
“The video is there, it gives you a visual of the person, but it's the audio that says, hey, I want you to wire me 25 million dollars. The intent happens through audio”
Four Design Partners, Zero Banks
The banking suite bundles three products that already existed on their own12. One app stores a person's verified credentials. Zoom's version runs the live video check. A third piece, World ID for Agents, handles a newer case. It lets a verified person hand those same credentials to an AI system, so the system can act for them. Four design partners are testing the bundle: OpenExchange, Bain Capital, VanEck Funds, and Outtake3. Each one trades, invests, staffs conference calls, or builds fraud software. Zero of them is a bank. World named zero banks as customers, and pricing stayed undisclosed.
World's bank-ready fix lands Oct. 18, 2026, after Arup's $25 million deepfake loss
Dated steps from the Arup deepfake loss to World ID for Zoom's general release
Sources: The Block [1]; AI Incident Database [4].
Read the chart as a table
| Date | Event |
|---|---|
| Feb. 2, 2024 | Arup wires $25 million to deepfake callers. A Hong Kong finance worker authorizes transfers after a video call with deepfakes of the CFO and colleagues |
| April 2026 | World ID for Zoom enters limited beta |
| Oct. 18, 2026 | World ID for Zoom reaches general release. World's banking suite launches the same day |
Scale is the pitch's real argument. Over 18 million people now hold a Proof of Human credential1. World counts verified users in more than 160 countries. The network has issued more than 475 million identity proofs across its partner platforms3. A design partner only tests a product, though. Wiring eight figures on a badge's say is a far bigger bet. Banks have placed zero such bets yet. Startup Fortune put the open question plainly. A bank's compliance department has to decide whether to trust a three-year-old crypto-adjacent identity company with wire-approval authority.
The Case for Watching the Face
Herbig's argument rests on scale and timing. Eighteen million Proof of Human credentials give World a large verified population already1. A bank's own customers and employees are likely inside it. Deloitte's own number backs the urgency. Generative AI could push US fraud losses to $40 billion by 2027. That figure triples the $12.3 billion Deloitte measured in 2023.
Deloitte expects US AI-enabled fraud losses to triple to $40 billion by 2027
Deloitte's 2023 figure versus its 2027 projection for US AI-enabled fraud losses, $ billions
- 2027 (projected): $40B
- 2023: $12.3B
Source: The Block [1].
Read the chart as a table
| Item | Value |
|---|---|
| 2023 | $12.3B |
| 2027 (projected) | $40B |
“World ID for banking and finance offers a suite of tools to add an AI-safe trust layer across their internal and external systems,” Herbig said, describing the same pitch Blania made onstage1.
World's own record gives Herbig's pitch a track record to point to. Sam Altman leads OpenAI as its chief executive13. He co-founded World three years ago. The company has since signed Tinder and Docusign to the same verification layer. Tinder uses it to confirm matches. Docusign uses it to confirm signers. A badge that already screens dating profiles and legal signatures is a reasonable starting point for a wire-approval call, the argument runs.
The Case Against Trusting the Tile
Vijay Balasubramaniyan runs Pindrop, a rival identity-security firm that sells its own deepfake detection to banks5. He locates the weakness in World's pitch at the exact moment the Arup fraud turned criminal4. “The video is there, it gives you a visual of the person, but it's the audio that says, hey, I want you to wire me 25 million dollars. The intent happens through audio,” he said. His point cuts at the tool's design. World ID for Zoom checks a face against a face, three images at a time1. The instruction that moves the money travels as sound.
Balasubramaniyan's own firm competes for the business World is chasing. His skepticism carries a commercial stake, then. It still lines up with the Arup record, too. The deepfakes on that 2024 call fooled the worker's eyes4. Its trigger line arrived as sound, well past the camera. A video-only check leaves that channel unchecked. Balasubramaniyan has described the broader scale elsewhere, too. Some of Pindrop's customers now field a deepfake attempt every three hours5.
World's own product chief has answered a different objection, one about privacy rather than efficacy6. Kenya suspended World's iris-scanning Orbs in 2023. Spain and Portugal ordered the same data collection stopped in 2024. Indonesia froze World's permit in May 2025. Each cited consent and data-collection rules. Tiago Sada argues back. The Orb builds zero central databases, he says. Its code also stays open for anyone to audit, he adds. He compares the early suspicion to the first reactions people had to Apple's Touch ID. A bank weighing World ID for its wire desk now faces two separate doubts. Does the check work? Will regulators let it run?
By the numbers
- $25 million: what an Arup worker in Hong Kong wired to fraudsters behind a deepfake video call on Feb. 2, 20244.
- 4: design partners testing World's banking suite; zero banks named as live customers13.
- $40 billion: Deloitte's projection for US AI-enabled fraud losses by 20271.
- 18 million: people who hold World's Proof of Human credential as of Oct. 18, 20261.
- $12.3 billion: Deloitte's measured US AI-enabled fraud figure for 2023, the baseline for that projection1.
- April 2026: when World ID for Zoom entered a limited beta, ahead of this week's general release1.
- 160-plus: countries where World counts verified users1.
- 475 million-plus: identity proofs World has issued across its partner platforms3.
What to watch
Watch for the first bank willing to put its name on a World ID deployment. Bain Capital, VanEck, OpenExchange, and Outtake still carry only the design-partner label13. Tools for Humanity's next real test looks different, though. Will it add a voice or audio check to answer Balasubramaniyan's objection directly5? The Arup instruction traveled by sound, after all4. Kenya, Spain, Portugal, and Indonesia are the governments worth tracking too6. Any reversal of their Orb restrictions would tell banks more about regulatory risk than another design partner ever could.
Sources
- Zack Abrams, "World launches World ID suite for banks, makes Zoom deepfake check generally available," The Block, Oct. 18, 2026, https://www.theblock.co/news/ecosystems/2026-10-18-world-launches-world-id-suite-banks-418869
- "World Pitches Banks a Bundle of Human-Verification Tools to Counter Deepfakes," Unchained, Oct. 18, 2026, https://unchainedcrypto.com/world-pitches-banks-a-bundle-of-human-verification-tools-to-counter-deepfakes/
- Walter Schulze, "Banks Add Sam Altman's World ID Badge to Calls to Spot Deepfakes," Startup Fortune, Oct. 19, 2026, https://startupfortune.com/banks-add-sam-altmans-world-id-badge-to-calls-to-spot-deepfakes/
- "Incident 634: Alleged Deepfake CFO Scam Reportedly Costs Multinational Engineering Firm Arup $25 Million," AI Incident Database, Feb. 2, 2024, https://incidentdatabase.ai/cite/634/
- Joel R. McConvey, "Biometric Update podcast digs into deepfakes with Pindrop CEO," Biometric Update, April 18, 2025, https://www.biometricupdate.com/202504/biometric-update-podcast-digs-into-deepfakes-with-pindrop-ceo
- Anush Jafer, "World Launches World ID Suite to Help Banks Tackle Deepfake Threats," Cryptopolitan, Oct. 19, 2026, https://www.cryptopolitan.com/world-id-suite-banks-deepfake-threats/