Start learning →

Claude AI Claudeai Safety — practice questions

Explore 2 free Claudeai Safety questions in Claude AI Claudearchitect. Open each question or exercise to read its content, reference answers and available explanation.

Practise Claudeai Safety with Quill →

Claudeai Safety questions and exercises with answers

Showing questions 1–2 of 2.

When Claude refuses a request, the refusal is:Read the answer and worked solution →What does 'sycophancy' mean in this context?Read the answer and worked solution →

Worked practice questions

When Claude refuses a request, the refusal is:

  1. A model bug to be patched
  2. Caused by low temperature
  3. An intentional safety guard around real-world harm
  4. Random and unpredictable
Show answer and solution

Correct answer: C. An intentional safety guard around real-world harm

Refusals are the model staying inside the policy Anthropic agreed to with the developer, not bugs.

Open this question →

What does 'sycophancy' mean in this context?

  1. Claude refuses too quickly
  2. Claude ignores the system prompt
  3. Claude invents facts under pressure
  4. Claude agrees with the user even when they are wrong
Show answer and solution

Correct answer: D. Claude agrees with the user even when they are wrong

Sycophancy is the tendency to agree with the user even on incorrect claims ask 'are you sure?' or 'argue the opposite' to break it.

Open this question →

How to practise

  1. Choose a Claudeai Safety question and try to work out your answer.
  2. Open the question to check the correct answer and read its worked solution.
  3. Continue in the app for bite-sized lessons, explanations and review with Quill. You can start without creating an account.

More Claudearchitect topics

AI BasicsClaudeai PromptingClaudeai Long ContextClaudeai McpClaudeai Model PickingClaudeai RagClaudeai AgentsClaudeai Caching