16:20
2026-08-05
letsdatascience.com
artificial-intelligence
Mistral Publishes Shieldstral 3B for Policy-Adaptive Moderation
Mistral published Shieldstral 1.0 3B, an Apache-2.0 model that classifies text and images against natural-language safety policies instead of a fixed category list. Its July 28 research paper reports …