Skip to content

AI Safety & Governance interview questions

Data privacy, PII masking, guardrails (NeMo, Llama Guard), alignment tax, copyright, and responsible AI deployment.

What to study

  • Implementing input/output guardrails using Llama Guard or NeMo.
  • Redacting PII and enforcing data isolation in multi-tenant RAG systems.
  • Managing the alignment tax: balancing safety interventions with model capability.

Questions and answers

  1. How do input and output guardrails prevent jailbreaks and data leaks?

    Intermediate · 4 min read

  2. What is prompt injection and how does it differ from SQL injection?

    Beginner · 2 min read

  3. How do you redact PII (Personally Identifiable Information) before sending prompts to LLMs?

    Beginner · 4 min read

  4. How does Llama Guard classify unsafe inputs and outputs?

    Intermediate · 2 min read

  5. What is the alignment tax and how does it impact model reasoning?

    Advanced · 2 min read

  6. What is system prompt exfiltration and how to prevent it?

    Beginner · 3 min read

  7. What is data exfiltration via LLM tool calls?

    Beginner · 2 min read

  8. What is hallucination and what are its primary causes?

    Beginner · 2 min read

  9. How do NeMo Guardrails enforce programmable rails using Colang?

    Intermediate · 2 min read

  10. How do you enforce strict RBAC and data isolation in enterprise RAG?

    Advanced · 2 min read

  11. What is indirect prompt injection and how does it occur when parsing web pages/documents?

    Beginner · 2 min read

  12. What is PII masking (pseudonymization) and how do Presidio/Regex filters protect user privacy?

    Beginner · 3 min read

  13. What is copyright infringement risk in RAG and fine-tuning datasets?

    Beginner · 3 min read

  14. What is model jailbreaking and how do safety classifiers block it?

    Beginner · 2 min read

  15. What is data lineage tracking for RAG documents and enterprise vector stores?

    Beginner · 2 min read

  16. How to prevent Data Exfiltration via Markdown image tags and hidden web beacons in LLM outputs?

    Intermediate · 4 min read

  17. How does Llama Guard taxonomy classify unsafe inputs and outputs across safety categories?

    Intermediate · 4 min read

  18. How to enforce Role-Based Access Control (RBAC) filtering in multi-tenant RAG vector search?

    Intermediate · 3 min read

  19. How do NeMo Guardrails use Colang state flows to strictly control conversation trajectories?

    Advanced · 3 min read

  20. How to audit and secure Model Context Protocol (MCP) servers against unauthorized tool invocation?

    Advanced · 4 min read

  21. What is model inversion and membership inference, and how do you defend against them?

    Advanced · 4 min read

  22. What is training data poisoning and how do backdoor triggers survive fine-tuning?

    Advanced · 4 min read

  23. What is model supply chain security and why are serialized model weights dangerous?

    Intermediate · 4 min read

  24. How does the EU AI Act classify AI systems into risk tiers?

    Intermediate · 4 min read

  25. What is the NIST AI Risk Management Framework and how do its four functions structure AI governance?

    Beginner · 4 min read

Study AI Safety & Governance with progress tracking