Nvidia Corporation Forms AI Safety Team for Open-Weight Models

Nvidia Corporation is assembling a new AI safety and security engineering team focused on evaluating AI agents before deployment and creating AI-powered tools to patch software vulnerabilities. This initiative supports the company's strategic push for open-weight models, which it argues promote transparency and strengthen American cybersecurity defense and AI leadership.

As part of this ecosystem, the company became a founding member of the Open Secure AI Alliance. This consortium, comprising 120 companies including Microsoft and SpaceX, is dedicated to the development of open-source security tools. CEO Jensen Huang has urged U.S. policymakers to support open models, claiming they enhance overall safety.

From a business perspective, Nvidia Corporation intends for this move to expand the customer base for open AI models. By fostering trust in autonomous AI agents for corporate environments, the company aims to drive higher demand for its AI chips.