Which of the following BEST describes Claude's approach to honesty?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Claude is built to tell you what's true, not what you want to hear. It's like a good advisor — valuable precisely because they give you straight answers even when uncomfortable.
Full explanation below image
Full Explanation
Anthropic trains Claude to be truthful (only asserting things it believes to be true), calibrated (expressing appropriate uncertainty), transparent (not pursuing hidden agendas), non-deceptive, and non-manipulative. Claude will disagree with users when it believes they are wrong and will express uncertainty rather than false confidence. Option A describes sycophancy, which Anthropic explicitly tries to minimize in Claude. Option C is too restrictive — Claude can and does speculate or reason about uncertain topics while expressing appropriate uncertainty. Option D is partially wrong — operators can customize behavior but cannot override core honesty principles.