Mozilla LLM Security & Cramming with Claude

December 5, 2025 · AIML@TU Final Meeting

Visit This Page

Scan to visit this page on your device

Attendance

Check in for today's meeting

Agenda

Today's Final Meeting

Guest Speaker: Luke Crouch on Mozilla LLM Security
Workshop: Cramming with Claude for Finals
Q&A and Closing Remarks

Guest Speaker

Luke Crouch, Privacy & Security Software Engineer at Mozilla

Luke Crouch

TU Alum & Privacy & Security Software Engineer at Mozilla Corporation

Presentation Focus

Internal research at Mozilla related to LLM security and privacy implications in web browsers

Join us as Luke shares insights from Mozilla's cutting-edge research on how Large Language Models interact with web security and user privacy.

Cramming with Claude

Using LLMs Effectively for Finals Preparation

Finals week is approaching! Learn how to leverage Claude and other LLMs as powerful study tools while maintaining academic integrity.

                        Workshop Topics
                        Prompt Engineering for Last-Minute Learning
Creating Custom Study Materials
Active Recall Techniques
Ethical AI Usage in Academics
Claude Code for Programming Practice

                    

Prompt Engineering for Last-Minute Learning

Strategic questioning techniques for deep understanding

The Feynman Technique Amplified

Use iterative questioning to drill down to true understanding:

"Explain [concept] to me, then say 'Now explain it simpler' or 'What's the intuition behind this?'"

Keep drilling until it clicks. You can also reverse it:

"I'll explain [concept] to you. Identify gaps in my understanding and ask me clarifying questions."

Targeted Concept Breakdown

Instead of asking "explain thermodynamics," try:

"What are the 3-5 core concepts in [topic] that everything else builds on?"
"What do students most commonly misunderstand about [concept]?"
"Connect [concept A] to [concept B]—how do they relate?"

The "Pretend I Have 2 Hours" Prompt

High-yield cramming strategy

When time is limited, get straight to the most important material:

"I have a final on [subject] in 2 hours. What are the highest-yield topics I should review, and what are common exam traps?"

                        Why This Works
                        Forces prioritization of core concepts
Identifies common pitfalls and misconceptions
Focuses on exam-relevant material
Helps you allocate limited time effectively

                    

Study Guide & Practice Problem Creation

Generate targeted practice materials

Practice Problem Generation

"Generate 10 practice problems on [topic], starting easy and increasing in difficulty"

"Create a mock exam for [course] with an answer key—but hide the answers until I ask"

Interactive Quizzing

"Quiz me on [chapter]. After each answer, tell me if I'm right and why"

This creates an interactive learning session with immediate feedback, helping reinforce correct understanding.

Active Recall Over Passive Reading

Transform materials into active learning tools

Why Active Recall Matters

Research shows active recall is significantly more effective than passive reading. Use Claude to create materials that force you to retrieve information from memory.

Custom Flashcards

"Turn these lecture notes into Anki-style flashcards with questions on the front and concise answers on the back"

Converts dense notes into spaced repetition format
Creates focused, testable knowledge chunks
Can be exported to flashcard apps

Worked Examples

Learn by watching, then doing

Step-by-Step Problem Solving

"Solve this problem step-by-step, explaining your reasoning at each stage. Then give me a similar problem to try myself."

                        Effective Learning Pattern
                        Step 1: Watch Claude solve a problem with detailed explanations
Step 2: Attempt a similar problem independently
Step 3: Compare your approach with Claude's feedback
Step 4: Repeat with increasing difficulty

                    

This scaffolded approach builds confidence and reveals gaps in your problem-solving process.

Ethical & Effective Use

Using AI responsibly in academics

Tutor, Not Ghost-Writer

Use LLMs to understand concepts, not to produce work you submit as your own. The goal is learning, not shortcuts.

Verify Important Facts

LLMs can hallucinate—always cross-check critical information against reliable sources like textbooks, lecture notes, or academic references.

Know Your School's AI Policy

Many institutions now have explicit AI usage guidelines. Familiarize yourself with what's allowed and what's not in your courses.

General Rule: If you're using AI to help you learn and understand material, that's generally acceptable. If you're using it to avoid learning or to misrepresent your own work, that crosses the line.

Claude Code / Gemini CLI

For Computer Science Students

Interactive Coding Practice

Tools like Claude Code and Gemini CLI provide powerful ways to practice programming:

Debug Code Interactively

"Here's my code. I think it does [X], but I'm getting [error]. Help me understand what's wrong."

Explain your reasoning—this helps identify misunderstandings in your mental model.

Algorithm Walkthroughs

"Walk me through this algorithm like I'm tracing it by hand with sample input [X]"

Great for understanding sorting algorithms, graph traversals, dynamic programming, etc.

Generate Test Cases

"Generate edge cases and test inputs for this function to verify it handles [specific scenarios]"

Learn to think like a tester and improve code robustness.

Resources & Summary

Comprehensive guide to the topics covered

Key Takeaways

Use LLMs as tutors to deepen understanding, not as shortcuts
Employ targeted, specific prompts for better results
Prioritize active recall over passive reading
Always verify important facts and cross-reference sources
Follow your institution's academic integrity guidelines
Practice ethical AI usage in all academic work