Actu
VNC and Intel turn the PC into a secure GenAI machine
December 2024 by Marc Jacob
VNC and Intel are bringing generative AI to personal computers. The combination of Intel’s next-generation processors, the Intel OpenVINO toolkit, and the secure VNClagoon platform enables the confidential use of large language models (LLMs) on local PCs or notebooks – independent of proprietary AI vendor platforms.
One of the biggest challenges in using GenAI is ensuring the security and confidentiality of data. Language models are typically hosted on proprietary cloud platforms provided by AI vendors, where data is processed and often used to further train their models. For security-critical applications, such as those in critical infrastructure, this is unacceptable. Similarly, data-sensitive organizations and enterprises need alternative solutions to use GenAI productively and securely while maintaining full control over their data.
From the data center to your computer
Previously, with toolkits like Intel OpenVINO, LLMs could be deployed in an organization’s data center to run at high performance – a significant step toward greater data sovereignty in GenAI usage. However, these models were too large and resource-intensive for desktop PCs and notebooks. Now, by combining VNClagoon and OpenVINO with Intel’s powerful new Lunar Lake architecture, LLMs can be significantly compressed and run locally on Intel-powered devices in a secure sandbox environment. This means all data, documents, and models remain offline, with AI queries executed locally, ensuring complete privacy.
Secure AI on the move
This advancement opens up entirely new possibilities for mobile GenAI applications, such as securely processing confidential data in the field. Local AI is finally becoming a reality, providing a practical implementation of Confidential AI. This guarantees independence from unauthorized access through offline functionality without compromising performance. It enables specific data and lightweight models to be used in secure local environments for applications such as aid organizations, police, or military operations. It also simplifies compliance with security regulations, including the EU’s AI Act.
