Most open-source LLM safety tools focus on technical and operational risks, but enterprises need a layered approach combining tools with organizational processes to address governance, legal, and financial risks that tools alone cannot mitigate.
This paper analyzes 21 open-source AI safety tools by mapping their capabilities to a standardized risk taxonomy with 32 categories.