VAST Data launches confidential AI system

VAST Data has introduced DataEnclave, a confidential AI runtime designed to secure the critical vulnerability in accelerator memory where AI models must be decrypted during inference. By utilizing hardware-isolated execution and strict attestation protocols, the system ensures that both sensitive customer data and proprietary model weights remain protected within controlled environments. This approach resolves the security gap that previously existed despite encryption of data at rest and in transit, offering a robust solution for enterprises and government agencies handling highly regulated or air-gapped data. The platform distinguishes itself by maintaining independent key control, allowing customers to retain ownership of their data keys while model builders secure their intellectual property. This dual-control mechanism supports fully sovereign deployments and generates a tamper-proof audit trail for all enclave actions. By treating model weights as a managed resource similar to data, organizations can enforce precise governance policies, ensuring that AI agents and models operate within defined security boundaries without exposing proprietary intelligence to unauthorized access. This development is highly relevant to open data initiatives as it addresses the fundamental tension between data utility and security. As organizations increasingly adopt open data principles, they must also ensure that collaborative AI environments do not compromise privacy or intellectual property rights. DataEnclave demonstrates how open data ecosystems can integrate confidential computing to foster trust, enabling the safe sharing and processing of information across diverse stakeholders while preserving compliance and sovereignty in an era of advancing AI technologies.

Source: manilatimes.net
Published on 2026-09-27

Tras auditorías de Contraloría e informes de Procuraduría, JEP dice que no hay corrupción »

Open Data

Related news