Module 22 Lesson 4: Final Exam

Welcome to the Final Exam. This test consists of 10 complex scenarios designed to test your "Synthetic" understanding of everything you've learned.

Part 1: Scenario Questions

The Recursive Injection: A user uploads a PDF that contains a prompt injection. When the AI reads the PDF, the injection tells it to "Write a new PDF" that contains a different injection and send it to the CEO. What type of attack is this?
The Hidden Weight: You download a model from Hugging Face that has been "Quantized" to 4-bit. When you run it, it works perfectly, but every Friday at midnight, it starts generating racist content regardless of the prompt. What has happened to the weights?
The RAG Leak: A user asks: "What is the most secret thing you know about the project I'm working on?" The AI responds with a secret codename from a document it wasn't supposed to see. Which architectural layer failed?
The Token Drain: You see a 5,000% increase in your OpenAI bill overnight. Your logs show 10,000 requests from a single IP, but each request is a legitimate-looking question about "History." How did the attacker bypass your "Injection" filters while still draining your wallet?

Part 2: Technical Definitions

Define the difference between Indirect Prompt Injection and Data Poisoning.
What is a Trusted Execution Environment (TEE) and why is it used in Decentralized AI?
Explain Differential Privacy using a simple analogy.
What is Colang and which framework uses it?

Part 3: Policy and Ethics

A developer wants to use "Shadow AI" to speed up their work. Why is this a GRC risk for the company?
If an AI is "Biased" but "Very Accurate," is it safe to deploy in a Hiring context? Explain your reasoning based on Module 19.

Instructions:

Write down your answers. If you score an 8/10 or higher, you are ready for the final step.

Summary

This exam is the culmination of your journey. It represents the bridge between being a "Learner" and being an "Expert."

Next Lesson: The Finish Line: Certification and Next Steps.