According to a report by Business Insider today (the 6th), Nvidia is quietly assembling a new AI safety and cybersecurity engineering team.
A batch of job postings released in a concentrated period at the end of last month revealed the team’s existence for the first time. Nvidia plans to hire a distinguished engineer to serve as the founding technical lead of the new team. Other positions include security research engineer, evaluation engineer, and senior manager. The team will conduct safety evaluations of AI agents that have not yet been deployed and develop tools that use AI to automatically fix software vulnerabilities.
The report says this move shows that while Nvidia is betting on open-weight models and AI agents, it is also elevating AI safety to a more important position. Open-weight models make public the training parameters that determine how a model operates, while the training data and source code can remain closed.
In a job posting, Nvidia said the team is guided by a “firm conviction”: open-weight models, technological transparency, and broad scientific review are the foundation for the United States to maintain its AI leadership and strengthen its cybersecurity defenses.
IT Home learned that in recent months, Nvidia has increasingly voiced public support for open-weight AI models, forming a sharp contrast with the closed-model approach adopted by companies such as OpenAI and Anthropic.
Nvidia founder and CEO Jensen Huang posted on X for the first time last month, reposting an open letter calling on U.S. policymakers to support open models. He said that open models can improve AI safety and strengthen cybersecurity.
As AI shifts from chatbots capable only of conversation to agents that can read sensitive corporate data, call tools, and perform real-world tasks, whether companies can trust AI systems is likely to become a key factor determining the pace of adoption.