Securing Enterprise GenAI: The Invisible Risk in RAG Pipelines and AI Agents

Summary

Nasscom published an article highlighting the security risks inherent in enterprise generative AI deployments, specifically focusing on RAG (Retrieval-Augmented Generation) pipelines and AI agents. The piece draws attention to attack surfaces that are often overlooked in production AI systems, including prompt injection through retrieved documents, data poisoning in vector stores, and unauthorized data access via agentic workflows that interact with enterprise systems. The article underscores t