Analysis · curated 30 Jun 2026
Prompt Injection Is Not a Chatbot Problem: How the Attack Surface Changes When Your LLM Has Tools
First reported agentsafelabs.com
Coverage timeline
Single-source analysis — first reported, latest, and curated coincide.
Why it matters
Defenders building agent security need to understand that monitoring the user input channel alone leaves numerous unmonitored injection vectors across tools, memory, and integrations.
An explainer arguing that chatbot-era prompt injection defenses (input filtering, output monitoring, instruction hierarchy, adversarial training) fail against agentic systems because injection enters through tools, retrieved documents, memory, API responses, and inter-agent instructions rather than the user message. It frames the issue as a system-architecture and trust-boundary problem rather than a model-tuning one.