Wiz collaborates with NVIDIA to advance ML research for data classification

The blog outlines how Wiz’s research team leveraged NVIDIA NIM to deploy Llama 3 models for ML-based data classification within Wiz DSPM, aiming to better identify sensitive database fields and infer database purpose from names and schemas. They report improved performance and scalability (e.g., roughly 2x faster than running the model as-is and ~25% faster than a popular open-source framework, with 1,000 samples processed in ~1.4s vs. 1.8s and ~3s) and highlight ease of multi-model deployment. The piece focuses on operational benefits of AI-powered security and collaboration with NVIDIA rather than describing a specific security incident or threat activity.