Atharva Shingote

Semantic Consistency Drift Monitoring for Training-Free Adversarial Prompt Injection Detection in Agentic LLM Pipelines

Large language model (LLM) agents that autonomously retrieve external content and execute multi-step action plans are increasingly deployed in enterprise and safetycritical settings. This architecture exposes a critical attack surface: adversarial content embedded within retrieved documents can silently redirect agent...