July 22, 2025
At this year’s NVIDIA GTC, a team of researchers unveiled a groundbreaking approach to artificial intelligence: Decentralized Confidential Machine Learning (DCML).
Modern AI systems force you to pick your poison: ✅ Use powerful proprietary models — and surrender your data. ✅ Use open-source models — but with no real monetization path.
NEAR AI challenges this false choice with a bold question: Why not have both privacy and performance — plus a sustainable business model?
DCML, part of the NEAR AI vision, introduces a new architecture that blends Trusted Execution Environments (TEEs) with a unique cryptoeconomic mechanism: Proof of Response. This enables AI models to run:
🟣 Confidentially — your data stays encrypted and private 🟣 Verifiably — results can be cryptographically proven 🟣 Monetizably — developers earn without exposing their model or user data
This unlocks powerful new capabilities for NEAR AI:
All of this is made possible through a deep integration with Phala Network, whose Confidential VM handles the heavy lifting, making it easy for developers to plug into this future of privacy-first AI.
NEAR AI isn't just about technology — it's about reshaping the incentives and ownership models around artificial intelligence. Here’s why it matters:
⭐ AI models with built-in safety controls ⭐ Sustainable monetization without harvesting user data ⭐ Permissionless global access — open to anyone ⭐ Lower-cost, energy-efficient infrastructure
In short, NEAR AI makes it possible to build user-owned, privacy-protecting, and profit-sharing AI systems — without relying on centralized gatekeepers.
The DCML preprint is now available, offering a deeper dive into the architecture and vision behind NEAR AI. If you're a builder, researcher, or AI enthusiast, now’s the time to get involved.
Download Nightly and join NEAR — and be part of the AI revolution that puts people, not platforms, in control.
