← 🛡️ AI Security, Privacy & GovernanceNEXT IN AI SECURITY, PRIVACY & GOVERNANCEJailbreaks and Red-Teaming Taxonomy→
Core
Agent Security: Tool Poisoning, Memory Poisoning, Containment
Agent security covers threats that only exist once an LLM can call tools and act on their results: malicious tool or MCP responses, poisoned long-term memory, privilege escalation through tool misuse, and goal hijacking. The defense is runtime containment, least-privilege tools, kill-switches, and blast-radius limits, not better prompting. Applied AI interviews probe it because anyone shipping agents has to reason about what happens when an untrusted string steers a system that can spend money or delete data.
a free account unlocks the core curriculum tier · no card
TAUGHT IN THESE COURSES
This concept is a reference. These lessons put it in sequence, with what comes before and after it.
Agent Engineering·Trusting nothing the agent readsSign in13mWhy an agent's threat model is not your application'sEvery security model you have used assumes code decides and data is inert. An agent erases that line, because the text it reads is the same channel as the instructions it follows. This lesson replaces the assumption before the next three lessons build on it.Agent Engineering·Trusting nothing the agent readsSign in14mDeciding what the agent may do, not what it may sayOnce you accept that an injection will eventually succeed, the only thing that bounds the damage is what the agent was permitted to do at that moment. Identity, scoping, and the actions that need a second signature.Agent Engineering·Trusting nothing the agent readsSign in13mWhat a sandbox buys, and the exit everyone leaves openAgents that write code or drive a browser need somewhere to run that is not your machine. What isolation actually contains, why the network is the exit that matters, and how to decide the blast radius before the run rather than after.
RELATED CONCEPTS
PRACTICE THIS IN REAL QUESTIONS
AI Security, Privacy & GovernanceHow do tool-result and memory poisoning attacks compromise an AI agent, and how do you defend?→AI Security, Privacy & GovernanceAn MCP server you already approved changes a tool's description. What breaks, and how do you defend against it?→AI Security, Privacy & GovernanceYour agent calls tools on behalf of users. How do you design its identity, credentials, and authorization?→AI Security, Privacy & GovernanceA tool-using agent reads untrusted web content. How do you defend against prompt injection?→System Design for AI in ProductionDesign a text-to-SQL feature: let users ask questions in natural language over a real database.→RAG & Agent System DesignWhat is the Model Context Protocol (MCP), and how do you design good tools for an agent?→
COMPANIES THAT ASSUME THIS