The NEAR AI team has introduced NEAR AI Cloud and NEAR Private Chat, two new offerings aimed at leveraging AI’s capabilities while ensuring privacy and user control. These tools utilize confidential computing and cryptographic proof to allow users and developers to process sensitive data with AI, maintaining full oversight.
Illia Polosukhin, founder and CEO of NEAR AI, stated that the company is taking a significant step towards user-owned AI with these private, verifiable products. He emphasized that privacy and verifiability are essential for maximizing available AI context for users and businesses, secured by hardware and cryptography-based guarantees.
NEAR AI Cloud enables companies and developers to execute AI models within secure hardware environments, specifically employing Intel TDX and NVIDIA Confidential Computing. These Trusted Execution Environments (TEEs) isolate each request, preventing external access, including from NEAR and the hardware providers. Each request also generates a cryptographic proof, detailing its execution path and method, which can be independently verified. This mechanism offers users assurance regarding the secure and private processing of their data.
Marvin Tong, CEO of Phala Network, a confidential cloud infrastructure provider, remarked that this initiative brings confidential computing into production. He noted that Phala had awaited a system capable of proving privacy at scale for years, and NEAR is the first to deliver it.
NEAR Private Chat offers a secure, daily interface to prominent open-source AI models. It provides an encrypted and private method for asking questions, from financial planning to health advice, without concerns about conversations being logged, shared, or used for training future models. Polosukhin reiterated that everyone deserves access to powerful AI without compromising their privacy, stating that Private Chat demonstrates this possibility.
NEAR AI Cloud is currently operational in production with several launch partners. Brave, a privacy-focused web browser with over 100 million users, features an integrated AI assistant called Brave Leo. Brian Bondy, CTO and co-founder of Brave, highlighted that the only acceptable form of AI at Brave protects privacy at its core. He expressed excitement about collaborating with NEAR to provide verifiable privacy and transparency, enabling users to verify Brave Leo’s adherence to its privacy commitments.
Phala, a confidential compute cloud platform, is another launch partner. Marvin Tong, CEO, mentioned that NEAR makes privacy verifiable rather than merely contractual, which unlocks new real-world use cases. OpenMind, a robotics operating system for autonomous systems, is also partnering. Jan Liphardt, founder and CEO of OpenMind, stated that robots in homes, hospitals, and public infrastructure require secure data processing, and NEAR AI ensures sensitive data remains private in live environments.
Designed for developers, NEAR AI Cloud is compatible with the OpenAI API, simplifying the migration of existing applications. SDKs in Python and TypeScript automate encryption and key management. The system supports streaming, fine-tuning, and high-throughput performance with minimal latency, typically ranging from 5 to 10 percent.
This launch marks the initial phase of NEAR’s broader strategy to support user-owned AI. The platform plans to expand its capabilities to include encrypted memory, AI agents, and decentralized coordination infrastructure, all secured by the NEAR blockchain. Polosukhin concluded by stating that every AI workload, from a chat session to a national system, should run privately, provably, and with respect for the user.
