Ethics & Safety · Advanced
LLM Security for Engineers: Injection, Exfiltration & Sandboxing
Threat-model AI features: prompt injection, indirect injection through retrieved content, data exfiltration, and tool sandboxing.
- 55 min lesson
- advanced level
- Engineering Track
- Narrated presentation
What you’ll learn
- Direct vs. indirect prompt injection
- Data exfiltration via links and tool arguments
- Least privilege and excessive agency
- Sandboxing generated code execution
- Server-side validation of tool arguments
- Red-teaming and injection regression tests
Inside this lesson
A narrated, slide-by-slide presentation with 1 prediction exercise, 1 knowledge check, 4 videos. You answer as you go, and the platform checks your work.