Zoom is taking a new step to keep video meetings safe by partnering with World, Sam Altman’s human identity verification company, to help confirm that the people on a call are real humans, not AI-generated imposters.
It’s a response to a problem that’s moved from “sci-fi risk” to real-world losses. In one of the most widely discussed cases from early 2024, engineering firm Arup reportedly lost $25 million after an employee in Hong Kong approved multiple wire transfers during what looked like a normal video meeting with the company’s CFO and other colleagues. The twist: everyone on that call, except the employee, was allegedly a deepfake. A similar style of deepfake-assisted scam was reported at a multinational company in Singapore in 2025, reinforcing that these aren’t isolated experiments—they’re repeatable playbooks.
The financial impact is rising fast. Estimates suggest deepfake-enabled fraud drove more than $200 million in losses in just the first quarter of last year. Security industry reporting also points to the average loss per corporate deepfake incident now exceeding $500,000. Even if most everyday users never encounter a deepfake scam personally, businesses that approve payments, share sensitive data, or make high-stakes decisions over video calls are increasingly exposed.
Why can’t platforms just “detect deepfakes” the usual way? World argues that many existing defenses rely on analyzing individual video frames for signs of manipulation. That approach is starting to break down as AI video generation improves, making traditional frame-by-frame deepfake detection less dependable over time.
Instead, Zoom’s new integration with World introduces a more layered verification method using World ID Deep Face technology. The system aims to prove a meeting participant is a real person through a three-part match:
1) A signed image captured during registration using World’s Orb device
2) A real-time face scan from the participant’s own device
3) A live video frame visible to other meeting participants during the call
Only when all three signals match does the system confirm the person’s identity, at which point a “Verified Human” badge appears next to that participant’s name in the meeting. It’s a notable shift from “try to spot what’s fake” toward “prove what’s real.”
Zoom says hosts will be able to turn on a Deep Face waiting room, requiring participants to verify before joining. Participants can also request verification mid-call, which could be useful if something feels off during a conversation—such as an unexpected request for money transfers, account changes, or sensitive documents.
According to Zoom, this fits into its broader approach of offering an open ecosystem so customers can add trust and identity tools that match their needs, especially in workflows where knowing who you’re speaking with is critical.
The partnership also reflects World’s growing push to bring human verification to more everyday digital experiences. In addition to video meetings, the company has been working with consumer-facing platforms such as dating and payments, and it recently introduced technology designed to confirm that real humans—not automated AI programs—are behind AI shopping agents at the moment of purchase.
As deepfake scams evolve from novelty to a scalable criminal tool, the ability to verify human identity in real time may become a standard requirement for business video conferencing—especially anywhere a call can trigger high-value financial actions.
