Nvidia Rolls Out AI Security Tools It Says Would Have Stopped Hugging Face Hack

Nvidia released new AI safety and security software designed to prevent breaches of AI agents, the automated systems that act on behalf of users inside AI applications. The company said the tools could have stopped the hack that hit Hugging Face, a popular platform for sharing AI models, though it did not detail how the incident occurred or how the new software would have blocked it. The launch is part of a broader push by Nvidia into AI security infrastructure, an area that has drawn more attention as companies deploy AI agents with greater autonomy across corporate systems. Nvidia is positioning itself as a supplier not just of the chips that run AI models but of the software layer meant to keep them safe. It is too early to know how widely enterprises will adopt the new tools or whether they meaningfully reduce breach risk in practice. Nvidia has not disclosed adoption numbers or named early customers for the software.