Nvidia is reportedly assembling a dedicated AI safety and security engineering team, as per a series of corporate job listings posted late last month. The hiring initiative signals a broader effort by the chipmaker to address security vulnerabilities and safety risks associated with autonomous AI agents, which are currently seen as one of the biggest threats of technology, and open-weight models before they reach widespread commercial deployment.According to the disclosures (spotted by Business Insider), Nvidia is recruiting an engineer to serve as the “founding technical leader” for the new team, alongside roles for a security research engineer, an evaluation engineer and a senior engineering manager.
Focus on autonomous agents and software patching
The primary responsibility of the newly-formed division is said to be evaluating autonomous AI agents prior to deployment, testing their operational limits and developing AI-powered security software to automatically detect and patch code vulnerabilities.One job description notes that the team’s mission is “rooted in the firm belief that open-weight models, transparency, and broad scientific scrutiny are foundational to American AI leadership and cybersecurity defense.”Unlike “closed” proprietary software, open-weight models make their trained parameters publicly accessible for inspection and fine-tuning, allowing outside developers and researchers to evaluate how the software operates.
Nvidia advocates for open-source security
The internal safety push aligns with Nvidia’s broader public strategy advocating for open-weight AI development, positioning the company alongside alternative approaches to closed ecosystems built by competitors like OpenAI and Anthropic.In a recent public letter addressed to US policymakers, Nvidia CEO Jensen Huang urged government support for open AI models, arguing that open architectures strengthen overall software resilience and national cybersecurity.Additionally, Nvidia recently became a founding member of the Open Secure AI Alliance, which is a consortium of 120 companies, including Microsoft, Palantir, SpaceX, and Hugging Face, focused on developing open-source security defenses for machine learning applications.