AI and LLMs · Senior · Question 29 of 75
What does defence in depth mean for LLM safety?
- 1Independent layers: input checks, rules, output checks, permissions
- 2Choosing the largest model, since it is the safest one
- 3One very long system prompt that lists every forbidden action
Show the answer
Correct answer: Independent layers: input checks, rules, output checks, permissions
Explanation
Each layer fails sometimes; together they make a full failure much less likely.