Accessibility
These controls work on your own device. Nothing is sent anywhere and nothing is kept. Calm screens always: nothing on this page flashes, strobes, moves on its own, or changes brightness suddenly.
Test — Certificate Topics — LLM Security — Protecting Generative AI Systems (Year 1 Foundations)
First, your location — certificates are counted by location. 10 questions. Answer at least 8 of the 10 correctly to pass (80%); a perfect score earns this lesson’s coin. Answers are never shown.
Question 1. Why are LLMs uniquely open to prompt injection?
Question 2. An indirect prompt injection is one where…
Question 3. The most sensitive question about a consumer AI tier is…
Question 4. A fluent, confident, invented citation in model output is…
Question 5. Excessive agency means…
Question 6. The unifying defense across the OWASP LLM risks is…
Question 7. Document-level access control matters in RAG because…
Question 8. Data minimisation, applied to an AI tool, means…
Question 9. NIST's AI Risk Management Framework is structured around…
Question 10. An acceptable-use policy for AI should, per the lesson…
Not an accredited school. Certificates of Completion only — no degree, diploma, or job promised. Not legal, medical or professional advice.