World Targets Banks With Proof-of-Human ID Tools, Citing $40B AI Fraud Risk

$40 billion. That is the AI fraud exposure figure Sam Altman's World is using to pitch its iris-scan identity verification technology to banks and financial institutions, marking a significant pivot from the project's troubled consumer-facing origins.

World announced this week that it is bringing its proof-of-human verification system into banking and financial workplaces, with Bain Capital, VanEck, and Pantera Capital signed on as design partners for the offering. The product also introduces a mechanism that lets users delegate verification proofs to AI agents, extending the system's reach beyond individual human authentication into automated workflows.

The expansion includes integration with Zoom meetings and financial workplace environments, positioning World's biometric stack as a deepfake prevention layer for institutions increasingly exposed to AI-generated fraud. Deepfake attacks, where synthetic audio or video is used to impersonate executives or clients, have become a documented threat vector in corporate finance. A finance employee at a multinational firm in Hong Kong was tricked into transferring $25 million in early 2024 after attackers used deepfake video to impersonate the company's CFO on a conference call, a case that drew widespread attention to the vulnerability.

The $40 billion fraud risk figure cited by World has not been independently verified, and the company has not disclosed the methodology behind the estimate. That number is doing significant marketing work here. Banks evaluating the product will want to scrutinize the sourcing before it becomes a budget justification.

World's underlying technology relies on iris scanning via its Orb hardware device to generate a cryptographic proof of unique human identity, stored on a blockchain. The system was originally called Worldcoin before a rebrand and launched in 2023 amid immediate regulatory friction. Authorities in Kenya suspended operations. Spain, Portugal, and Germany opened investigations into the data collection practices. The core tension has always been the same: to prove you are human, you hand over one of the most sensitive biometric identifiers you have.

That history complicates the banking pitch. Financial regulators in major markets operate under strict data protection frameworks, including GDPR in Europe and a patchwork of state-level biometric privacy laws in the United States. Whether World's architecture, even with zero-knowledge proofs designed to avoid storing raw iris data, satisfies those frameworks at the institutional level remains an open question. Banks tend to move slowly on unproven identity infrastructure, particularly when the vendor carries regulatory baggage across multiple jurisdictions.

The AI agent delegation feature adds another layer of complexity. The stated purpose of a proof-of-human system is to confirm a living, consenting individual is present. Allowing that proof to be delegated to an autonomous agent reintroduces ambiguity into the very problem the system is meant to solve. Security researchers have flagged similar concerns about proxy authentication schemes, where a legitimate credential is issued once and then used by software operating without ongoing human oversight.

WLD, the project's native token, is trading roughly 96% below its all-time high, a gap that reflects persistent market skepticism about the project's trajectory. Token price is not a direct measure of a protocol's technical merit, but in this case the underperformance tracks with a broader pattern: repeated regulatory setbacks, slow Orb deployment, and a user base that has grown more slowly than early projections implied.

The institutional pivot is a logical response to those headwinds. Consumer adoption of iris-scanning hardware faces obvious friction. Selling verification-as-a-service to banks, which already operate compliance infrastructure and have clear financial incentives to reduce fraud losses, is a more tractable market. Bain Capital, VanEck, and Pantera's involvement as design partners rather than investors signals a product development relationship, lending some credibility to the seriousness of the banking push without guaranteeing commercial uptake.

Whether financial institutions move beyond the design partner phase to actual deployment will depend on answers that World has not yet provided publicly: which regulators have signed off, how biometric data is handled under existing banking privacy rules, and what audit trail exists when an AI agent presents a delegated human proof. Until those questions have clear answers, the $40 billion fraud risk figure is a conversation starter, not a closed deal.