Interview, Fireside Chat
How Bots, Deepfakes and AI Agents Are Forcing a New Internet Identity Layer | Alex Blania on a16z
- Current visible bot activity is less than 1% of the expected scale, with the full impact of superlinear agentic capability increases and dropping intelligence costs projected within the next year or two.
- AIs will achieve human-like behavior, pass the Turing test, replicate the "web of trust," and manipulate human psychology to a degree where distinguishing humans from agents becomes nearly impossible on video and social platforms.
- Biometric solutions will shift as Face ID and fingerprints hit mathematical limits for global uniqueness, making iris scanning the only viable modality for massive-scale verification, potentially normalized through AR/VR devices like the Apple Vision Pro.
- Companies plan to deploy approximately 50,000 Orb devices across the US to achieve an average access time under 15 minutes, while launching "Orb On Demand" to deliver verification to users in the Bay Area or New York within 50 minutes via motorbike.
- Immediate strategies include deploying temporary Face Check solutions for rate limiting (capping accounts at 10-20 instead of 100), seeking partnerships with large entities like Walmart and Starbucks, and targeting DMVs to normalize technology, all before full platform integration.
- Regulatory and infrastructure plans rely on the anticipated US "Clarity Act," cryptographic infrastructure to prevent fraud in social security and Medicare, and zero-knowledge proofs to maintain user anonymity while verifying unique human status.
- Market risks have shifted from thesis validation to execution challenges regarding the distribution of billions of devices and economic feasibility, with competitors expected to attempt similar solutions roughly six years behind current efforts.
- Business models for video platforms and advertising face disruption as AI-generated content becomes highly scalable and bots watch AI content, while creators may lose audience support if interactions are detected as non-human.
- The company has verified 18 million users and holds 40 million in the app, with 90% of future effort focused on the US market to build a valuable human network against AI-driven impersonation.
- Deep fakes will become photorealistic and real-time within a year, necessitating mandatory human verification for high-value applications like video conferencing for fund managers, dating apps, and political processes to preserve democracy.