NVIDIA leads new AI security alliance after Hugging Face breach

NVIDIA, alongside Microsoft and 25 other founding members, has launched the Open Secure AI Alliance to enhance cybersecurity following the Hugging Face AI incident. The coalition’s goal is to utilize open technologies to address and disclose vulnerabilities in AI systems. The alliance's formation was triggered by a breach at Hugging Face, where an unknown AI agent attacked, highlighting vulnerabilities in AI model security.

Each member contributes specific expertise, such as NVIDIA providing open models and Hugging Face offering secure formats for model weights. The alliance calls for broader collaboration among governments and enterprises, advocating that open AI models should be seen as defensive assets to strengthen cybersecurity.