CCAR-P: Claude Certified Architect — Professional

By Anthropic · Question Mode

QuestionQ88

VerifiedGovernance, Safety & Risk Management

A Claude architect is implementing safety controls for a customer-facing advice assistant that must never give regulated investment recommendations.

Which two guardrail implementations most directly enforce this requirement? (Select two.)

Choose two
Explanation

An output classifier can block prohibited investment-recommendation language before it is delivered, while explicit system-prompt boundaries and prescribed refusal wording direct the assistant to decline out-of-scope investment-advice requests. Anthropic guidance recommends strengthened system prompts that state legal and ethical boundaries and how to refuse, and describes filters as a way to enforce output guardrails.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!