Loading…

Share feedback

How are you feeling about Velocode?

Or email us at hello@velocode.ai

Learn/AI Security & Prompt Injection

Course

AI Security & Prompt Injection

The AI-security surface: direct and indirect prompt injection, jailbreak defense, red teaming, guardrail architecture, and layered runtime defenses.

45 practice questions32 free13 company-verified (Pro)

Start the course →Jump to practice

Course chapters

1
The AI attack surface
14min · concept, worked example, practice & checkpoint
2
Prompt injection
16min · concept, worked example, practice & checkpoint
3
Jailbreaks & red teaming
14min · concept, worked example, practice & checkpoint
4
Guardrails & output validation
14min · concept, worked example, practice & checkpoint
5
Data privacy & governance
13min · concept, worked example, practice & checkpoint

Free to practice

Role-Play Jailbreak Incident Response

AI Securityeasy

Indirect Prompt Injection Against an AI Research AssistantAI Securityeasy

Defending the Argument that the Model Itself is an Attack SurfaceAI Securityeasy

Threat Model for a Fintech Customer Service ChatbotAI Securityeasy

LLM01: Prompt InjectionAI Securityeasy

Prompt injection attacks compromise AI system integrity through adversarial inputsAI Securityeasy

LLM Jailbreaking Techniques: Bypassing Safety Guidelines EffectivelyAI Securityeasy

Prompt Injection Attacks: Security Vulnerabilities in LLM-Based ApplicationsAI Securityeasy

LLM Jailbreaking: Bypassing Safety Constraints Through Prompt EngineeringAI Securityeasy

Customer PII Is Leaking Into the LogsForward Deployedmedium

Support Agent Leaks Data via Prompt InjectionForward Deployedmedium

Agent Security — Exfiltration Attacks on a Coding AssistantAI Securitymedium

RAG Security — Retrieval Poisoning in a Financial AI AdvisorAI Securitymedium

AI Agent Design — ReAct Loop and Tool SetAI Engineeringmedium

EU AI Act + NIST AI RMF Compliance for an AI Hiring SystemAI Securitymedium

Red-Teaming an AI Assistant for a Law FirmAI Securitymedium

Defence-in-Depth Architecture for a Cloud Infrastructure AgentAI Securitymedium

Prompt Injection Defense Techniques and Implementation StrategiesAI Securitymedium

Prompt Injection Attacks and Defense Mechanisms for LLM SystemsAI Securitymedium

Detect prompt injection attacks in user input stringsAI Securitymedium

Prompt Injection Defense Mechanisms and Mitigation StrategiesAI Securitymedium

Red Team Strategy Development for Pre-Launch LLM Chatbot SafetyAI Securitymedium

Healthcare Chatbot Safety: Implementing Diagnostic Guardrails and DisclaimersAI Securitymedium

LLM Output Guardrails Detecting Off-Topic Responses and PII LeakageAI Securitymedium

How do you design system prompts that are robust across diverse users?AI Engineeringmedium

LLM Output Guardrails: Topic Relevance and PII Detection SystemAI Securitymedium

One Customer Can See Another Customer's DataForward Deployedhard

Two Teams Want the Agent to Behave DifferentlyForward Deployedhard

Stop the Agent Inventing Facts in a Regulated DomainForward Deployedhard

Backdoor Detection — Assessing a Pre-Trained Text ClassifierAI Securityhard

Guardrails and Safety — Childrens Platform ArchitectureAI Engineeringhard

How do you choose a vector database for production RAG?AI Engineeringhard

Company-verified Pro

Real AI Security & Prompt Injection questions reported from interviews at 50+ AI companies, with the golden answer and full scoring. Unlock with Pro →

Zero trust security assumes no user or device is trusted by default.Reported at Amazoneasy
Access control filter placement before reranker causes leakage bugReported at Coheremedium
Adversarial Attacks on Large Language Models: Vulnerabilities and DefensesReported at Snapmedium
GenAI Safety: Building Trust Through Guardrails and EvaluationReported at Anthropicmedium
Generative AI safety: detecting and mitigating harmful content generationReported at Anthropicmedium
Design a prompt injection defence architectureReported at Anthropicmedium
Implement a RAG pipeline with citation trackingReported at OpenAImedium
Multi-tenant RAG: Secure Vector Index Isolation for EnterpriseReported at OpenAIhard
Red-team harness for automated prompt-injection bypass discoveryReported at OpenAIhard
Defending Proprietary LLM APIs Against Model ExtractionReported at Nvidiahard
Secure GPT-4o Agent Against Indirect MCP Prompt Injection AttacksReported at OpenAIhard
Bot Detection System Design for Gaming PlatformsReported at Robloxhard
Design a multi-tenant vector database for 10,000 enterprise customersReported at Google DeepMindhard

Ready to master AI Security & Prompt Injection?

Answer real questions in the browser, scored on correctness, architecture, token efficiency, and security against a golden answer.

Start practicing free →

Practice Learn Interview Pricing