Safety & Guardrails

90 questions

Deploying LLMs safely. Content filters, Llama Guard, NeMo Guardrails, prompt injection defense, jailbreak detection, red-teaming, OWASP LLM Top 10.

Premium questions for this topic

Sort

Questions

Contrast RLAIF and RLHF as paths to safer model behaviour
FlashcardMedium
Order the layers of a defence in depth safety stack from innermost to outermost
Order StepsMedium
Match PyRIT and garak to the red teaming jobs they are actually good at
Match PairsMedium
Explain what Microsoft PyRIT contributes to LLM red teaming
FlashcardMedium
Describe garak's role in the red team toolchain
FlashcardEasy
Identify what Lakera Gandalf is and its training data value
FlashcardEasy
PremiumPick where a continuous…
Multiple ChoiceMedium
PremiumPick the strongest reason…
Multiple ChoiceMedium
Identify the core problem weak to strong generalization tries to solve
Multiple ChoiceHard
Pick the most accurate summary of the Anthropic sleeper agents result
Multiple ChoiceHard
Distinguish deception from sandbagging in frontier model evals
FlashcardHard
Explain alignment faking and what its existence implies for RLHF
FlashcardHard
Explain how SR 11-7 applies to LLM use in a bank
FlashcardHard
Tell apart the NIST AI RMF from its Generative AI profile
FlashcardEasy
Match the four NIST AI RMF core functions to what they actually demand
Match PairsMedium
Describe the four NIST AI RMF functions and how the GenAI Profile extends them
FlashcardEasy
Identify what ISO/IEC 42001 certifies and why auditors increasingly cite it
FlashcardMedium
Match each compliance regime to the concrete guardrail control it forces
Match PairsMedium
Pick the EU AI Act obligation that most directly drives runtime guardrail design
Multiple ChoiceMedium
Match each OWASP LLM Top 10 risk to the control that addresses it
Match PairsMedium
How LLM10 model theft happens against a hosted API
FlashcardMedium
Describe LLM09 overreliance and one concrete UX mitigation
FlashcardEasy
PremiumPick the clearest example…
Multiple ChoiceMedium
Identify the insecure plugin design flaw (LLM07) in this snippet
Multiple ChoiceMedium
Walk through the three production paths through which LLM06 (sensitive disclosure) typically leaks
FlashcardMedium

Showing 1–25 of 90

← PrevNext →