Jensen Huang advises strict access controls for corporate AI adoption

Nvidia CEO Jensen Huang advised companies to implement strict security protocols when adopting artificial intelligence, comparing the integration of AI to hiring a new employee. During a keynote address at the Equinix Horizon event in San Francisco, Huang emphasized that just as new hires are not immediately granted access to all sensitive company files, AI systems must be assigned clear and restricted access permissions to prevent unauthorized data leaks.

Huang predicted that future AI will evolve into a ‘highly distributed’ structure, operating across various data centers and cloud environments rather than staying in a single location. This fragmentation necessitates robust security controls to protect information within decentralized networks.

Regarding the competition between closed and open-source AI models, Huang stated that both are essential. He noted that while cutting-edge models are necessary for high performance, open-source models are vital for areas requiring specific data control, security, and regulation.

In conjunction with the event, Nvidia and Equinix, alongside the startup Together AI, introduced the ‘Equinix Inference Exchange,’ an enterprise-grade distributed inference platform designed to securely connect fragmented data and distributed AI.