Become an AI engineer

LLM security

LLM security is a unit of 11 lessons in Rabbithole's Become an AI engineer course, at levels 3 to 5.

Lessons

1. Excessive Agency: when an LLM can do too muchLevel 3
2. Least privilege for agentsLevel 3
3. LLM Supply Chain SecurityLevel 3
4. OWASP Top 10 for LLM Applications: Output Handling, Unbounded Consumption, and Tool-Calling RiskLevel 3
5. Prompt injection: where the instruction comes from, and how CaMeL blocks the damageLevel 3
6. Sensitive Information DisclosureLevel 3
7. Data exfiltration: how it works and where the SSRF defences goLevel 4
8. Indirect prompt injection: hijacking a model through what it readsLevel 4
9. The Instruction HierarchyLevel 4
10. Safety classifiers: scoring a response while it streamsLevel 4
11. Adversarial attacks on aligned models: GCG and its defensesLevel 5

Sources

These lessons draw on 19 sources. The most used:

• OWASP Top 10 for LLM Applications 2025
• Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations (NIST AI 100-2 E2025)
• Lessons from Defending Gemini Against Indirect Prompt Injections
• The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions
• Defeating Prompt Injections by Design
• OpenAI Codex and ChatGPT desktop documentation (ChatGPT Learn)
• Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming
• Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection

All units in Become an AI engineer