AI Security: Prompt Injection, OWASP LLM Top 10 & Guardrails
Coming soon BeginnerIntermediateAdvanced
Get notified on YouTube Udemy certificate: coming soon
- Udemy
- Coming soon
- Lessons
- 15
- Planned length
- About 3 h 7 min
- Practicals
- Planned: a hands-on practical and a Udemy assignment in every section, plus a capstone
What you will be able to do
- Decide which actions need approval
- Block unsafe inputs/outputs and sandbox agent actions
- List the main attacks and redact PII before the model
- Reproduce and mitigate an injection attack
- Check an app against the current OWASP list
- Lock down an agent's tools and servers
Syllabus
AI Agents: Human-in-the-Loop and Guardrails
- Human-in-the-Loop Explained Decide which actions need approval
- Guardrails: Inputs, Outputs, Sandboxes, Least Privilege Block unsafe inputs/outputs and sandbox agent actions
LLM Engineering and Production: Security
- AI Security Basics and Data Leaks (PII, Secrets, Logs) List the main attacks and redact PII before the model
- Prompt Injection: Direct and Indirect Reproduce and mitigate an injection attack
- OWASP Top 10 for LLM Applications Check an app against the current OWASP list
- Agent and MCP Security: Excessive Agency, Tool Poisoning Lock down an agent's tools and servers
- Guardrail Tools: Llama Guard, NeMo Guardrails, Guardrails AI Add a safety classifier to an app
Frontier and Advanced: Agent Security and Red Teaming
- How Agents Get Attacked: The Risky Combination Spot the risky combination in any agent design
- Agent Defences That Hold Against Prompt Injection Apply design patterns that limit what an injected prompt can do
- MCP Security Lab: Tool Poisoning, Tokens, OAuth Scopes Attack and then harden a test MCP setup
Frontier and Advanced: AI Governance at Work
- Write an AI Use Policy for Your Team Draft a one-page policy: approved tools, data rules, review steps
- India DPDP Act Basics for AI Teams Map an AI feature to DPDP Act duties and the phased DPDP Rules 2025 timeline
- Privacy Engineering for AI Apps: PII and Retention Redact PII and set retention before data reaches a model
- AI Risk Management: NIST AI RMF and ISO/IEC 42001 Run a lightweight risk review for a new AI feature
- Audit Trails for AI Agents Keep the records you need when an agent gets something wrong
Take before
Lesson list from the ADITEJUS curriculum. Planned lengths are estimates per lesson; Udemy shows the final length.