How Bots, Deepfakes and AI Agents Are Forcing a New Internet Identity Layer | Alex Blania on a16z
Alex Blania explains why iris biometrics and multi-party computation are the only durable solution to bot-scale identity fraud as AI agents become indistinguishable from humans.
- World has 18M orb-verified users and 40M total app users; the company is now shifting 90% of focus to US market after avoiding it due to crypto regulatory uncertainty.
- University of Zurich study found AIs were superhuman at changing minds on Reddit’s Change My Mind subreddit by profiling users’ political motivations and communication styles.
- Face biometrics and web-of-trust approaches hit a wall at tens of millions of users due to insufficient mathematical entropy; iris is the only biometric with enough uniqueness for global 1-to-N matching.
- Iris codes are split via multi-party computation so no single party — including World — holds a complete biometric record; a zero-knowledge proof lets users prove uniqueness to platforms without revealing identity.
- Tinder launched World ID verification in Japan to badge real humans; video conferencing is the next high-value target because real-time deepfakes will be a commodity within a year.
- To cover the US, World estimates ~50,000 orb deployments needed to get average verification time under 15 minutes; distribution targets include Walmart, Starbucks, DMVs, and a forthcoming Orb on Demand motorbike service.
- $400B in COVID stimulus was estimated stolen partly due to no unique-human verification; Blania argues AI-scale fraud will make existing social security and Medicare systems functionally unworkable without cryptographic identity infrastructure.
2026-04-02 · Watch on YouTube